Innovative Statistical Model Uncover Effective Herbal Medicines Among Personalized Treatment Plans in Persian Medicine: A Small-Scale Study in Type 2 Diabetes

Abstract

Objectives:

In holistic medicine, developing personalized treatment plans is challenging due to the multitude of possible therapy combinations. This study introduces the use of a statistical approach to identify the most effective herbal medicines prescribed in Persian medicine (PM) in a small-scale sample of patients with type 2 diabetes mellitus (T2DM).

Methods:

This prospective observational cohort study was conducted with 47 patients with T2DM referred to Behesht Clinic in Tehran, Iran. A physician prescribed individualized PM treatment for T2DM and related systemic issues. The fasting blood sugar (FBS) level of each patient was recorded at initial and two follow-up visits, with visit intervals and treatment modifications determined by patient health status. Patients who completed two follow-up visits were included in the final analysis (n = 27). Data were analyzed using R software. A general linear model was assumed for the mean response, along with an exponential covariance pattern model, to manage irregularly timed measurements.

Results:

Two fitted models showed that, after adjusting for confounders, the use of the “Diabetes Capsule” significantly reduced the average FBS by 17.14 mmol/L (p = 0.046). For each unit increase in the consumption of “Diabetes Capsule” or “Hab-e-Amber Momiai,” the average FBS decreased by 15.22 mmol/L (p = 0.015) and 14.14 mmol/L (p = 0.047), respectively.

Conclusion:

It is possible to observe which medications are most effective, even when treatments are applied in a holistic and personalized fashion. Preliminary studies such as these may identify promising products for testing in clinical trials conducted under standardized conditions, to inform initial choices for future personalized treatments.

Introduction

Persian medicine (PM), also known as Iranian traditional medicine, is a holistic school of medicine with roots in ancient Persia. PM is based on a philosophical theory and categorical framework that was developed and expanded on by Persian scholars over centuries. The scientific basis of PM is demonstrated through influential texts such as Avicenna’s Canon of Medicine, a primary medical reference in Europe until the 16th century. As a long-standing, innovative, and scientifically grounded traditional medicine (TM), PM represents a valuable source of medicinal data and knowledge.¹

PM has recently seen a resurgence as a specialized academic discipline. To bring this ancient knowledge to the international community, it is vital to encourage in-depth research and discourse. This will ultimately contribute to the development of effective treatment strategies with a transnational reach.² Measuring the efficiency and benefits of traditional therapies is important for healthcare providers and health policymakers.³ There is therefore an urgent need to strengthen the evidence base of PM through the careful adoption of rigorous research methodologies. This is in line with the World Health Organization’s aim to advance scientific methods to better comprehend, evaluate, and apply more holistic, context-specific, complex, and personalized approaches to health and well-being that are emphasized in the 2023 Gujarat Declaration.⁴

Although the evidence-based medicine framework in Western medicine considers randomized controlled trials to be the best tests of effectiveness for healthcare interventions, it is unfortunate that many instances of PM cannot be explained well enough through these methods.³ Traditional clinical trials rely on a randomized controlled design, where participants are assigned to different treatment arms receiving fixed interventions throughout the study. Regular outcome measurements are performed on each arm, sometimes compared with a placebo to account for the placebo effect. In such studies, the strength of the evidence increases with increased similarity between the groups, assuming control for confounding factors. Balanced over time designs ensure comparable observations by maintaining consistency in the number and timing of measurements across groups. In essence, clinical trials strive for homogeneity—similar dosage, frequency of visits, and overall conditions—with interventions remaining unchanged until study completion.

In contrast to Western medicine, PM embraces a holistic and personalized approach, aligning with the principles of many complementary medicine systems, including Traditional Chinese Medicine. This orientation emphasizes dynamic, individual, and holistic perspectives during clinical practice Treatment plans are tailored to each patient’s specific condition, considering the overall health of their entire body system. Given that no two individuals, even with the same disease, share identical organ system functions, it is inherently impossible to prescribe identical treatments from the outset and maintain them throughout the study. The need for treatment (dosage and timing of visits) is dictated by a patient’s holistic condition, not solely by the disease name. TM emphasizes tailoring treatment to the specific needs of each patient, which can evolve throughout treatment. This inherent variability makes it practically impossible to achieve complete equivalence of treatment conditions.⁵

Therefore, there is a critical need to develop research methodologies specifically designed for complementary medicine studies. These methods must respect the core principles of holism and personalization while adhering to rigorous scientific standards and minimizing potential errors in analyzing patient responses to treatment.

To address these limitations of clinical trials in assessing personalized treatments, we proposed a prospective observational cohort study with an appropriate longitudinal modeling approach using a general linear model for the average response, coupled with an exponential covariance pattern model to handle irregularly scheduled measurements. By relaxing the balanced timing restriction, this strategy allowed researchers to collect and analyze repeated measurements of patients with personalized visit times according to their health and well-being status. Another important feature of the proposed statistical method is that the physician can determine the type and dose of drugs for each patient at each visit based on their special needs under the personalized medicine approach.⁶

Linear regression models are well-known statistical methods for the analysis of cause and effect relationships between a response or dependent variable (effect) and a set of predictor or independent variables (cause). Regression estimates provide (1) interpretation of the effects of significant risk factors on the mean response and (2) a prediction tool for the response of the new cases with unobserved responses. Two of the main assumptions in linear regression modeling are the independence of response measurements and the constancy of response variance among observations which are appropriate for modeling data from a cross-sectional study. However, when the study includes repeated measurements of the response variable for each individual, the independent assumption would be violated. Also, it might be necessary to assume different response variances in different measurement times. This calls for a more general linear model with the ability of modeling correlated responses with the possibility of heterogeneous variances for longitudinal data analysis.

The primary goal of a longitudinal study is to characterize the change in response over time and the factors that influence this change. Moreover, correlation and heterogeneous variability must be accounted for to obtain valid inferences about response change over time. Longitudinal data modeling requires more sophisticated statistical techniques because (1) repeated measures on the same individual are correlated and (2) variability is often heterogeneous across measurement occasions. Models for longitudinal data must jointly specify models for the mean response over time and the covariance among repeated measures. There are two broad classes of approaches for longitudinal modeling: marginal and mixed. This article proposes the application of a marginal approach to a general linear model framework. This approach to modeling longitudinal data permits individuals to be measured on different occasions and at different times while allowing a range of different covariance structures. Therefore, in this study, the mean response model was assumed to be a parametric curve against time, and an exponential covariance pattern was considered. The exponential pattern for covariance among repeated measures allows for unequally spaced measurement occasions over time. In addition, in this pattern, the correlation between any pair of repeated measures is assumed to decrease exponentially with time separation between them.^7,8

The model aims to assess the efficacy of various treatment options and pinpoint the most promising combination of treatments for patients receiving personalized care. By utilizing advanced statistical analysis techniques, the model can objectively determine which treatments yield the most favorable outcomes.

This approach helps to organize physician/s’ prescriptions. It focuses on one or a few drugs that have the best response, reducing the costs of noninsured herbal medicines and increasing patient satisfaction by receiving fewer drugs, faster response, and lower costs.

Therefore, this study aims to investigate the effectiveness of herbal medicines prescribed based on PM in patients with type 2 diabetes mellitus (T2DM) in a small-scale preliminary study using the proposed method.

Methodology

Study design

This study is a prospective observational single-center cohort study with an appropriate longitudinal modeling approach using a general linear model for the average response, coupled with an exponential covariance pattern model to handle unbalanced repeated measurement. Unlike clinical trials, patients were not grouped to receive the same prescriptions and their visit intervals were not necessarily equal. In this study, a physician (TM specialist) with a holistic approach not only treated diabetes but also addressed any related systemic issues. Simultaneously, the researcher recorded the fasting blood sugar (FBS) level of each patient in all three visits (including the initial and two follow-up visits). The researcher was solely an observer and did not influence the visit intervals, medication type, or dosage.

Participants

The study population (n = 47) consisted of patients with T2DM with no age or sex restrictions who were referred to Behesht Clinic affiliated with the Iran University of Medical Sciences in Tehran, between May 2021 and September 2021, with available FBS records (n = 47).

Patients who were diagnosed with T2DM by an endocrinologist were enrolled in the study after obtaining informed consent. The diagnosis of T2DM was based on several criteria: FBS levels between 100 and 125 mg/dL, random blood sugar levels of 200 mg/dL accompanied by polydipsia, weight loss, and polyuria, or 2-h post-challenge plasma glucose over 200 mg/dL. Patients who completed two follow-up visits were included in the final analysis (n = 27). Pregnant or breastfeeding women, individuals with severe medical conditions, a history of allergic reactions to herbs, uncontrolled or severe diabetes, and those with type 1 diabetes (T1DM) were excluded from the study. A flowchart depicting the study design and patient flow through each stage of the study is presented in Figure 1.

FIG. 1.

Study flowchart.

Medication information

The herbal medicines used in this study (Table 1) were produced by reputable Iranian pharmaceutical companies specializing in herbal remedies. The products have been approved by the Ministry of Health and Medical Education of Iran, and their Iran Registration Codes can be found at the Iran Food and Drug Administration in the following website: www.fda.gov.ir, with the exception of “Hab-e-Amber Momiai,” an Unani drug available at Persian medical clinics in Iran, which is manufactured by Hamdard Laboratories, a major pharmaceutical company based in Karachi, Pakistan.

Table 1.

Characteristics of Prescribed Drugs

No.	Prescribed drugs name	Drug ingredients	Dosage form	Daily dosage and route of administration	Indications^a	Company name/location
1	Diabetes Capsule	Bitter apple, tragacanth, Arabic gum, aniseed	Capsule	1–2 capsules a day with water	Reduce blood glucose level/Remove any thick or viscid matter	Talaye Sabz Tuba Co./Iran
2	Diabetes Powder	Nettle, fenugreek, tree germander, greater burdock, coriander, sagebrush	Powder	Decoction/1–3 times a day	Reduce blood glucose level	Talaye Sabz Tuba Co./Iran
3	Diabetes Oil	Bitter apple, squill, sesame oil	Topical oil	Apply the oil to the painful regions, palms and soles, extending to areas usually covered by gloves and socks	Reduce blood glucose level/Eliminate diabetic neuropathic pain	Talay e Sabz Tuba Co./Iran
4	Hab-e-Amber Momiai	Cinnamon, pistachio, clove, shatavari, bamboo, nutmeg, white turmeric, ginger, peony, mastic, amber, mummy	Tab	—One to two tablets a day with water	For sexual debility, tonifying vital organs	Hamdard Laboratories/Pakistan
5	Kabed Capsule	Fumitory, jujube, coriander, rhubarb, thyme, lettuce, chicory, senna	Capsule	Two to four capsules a day with water	Liver detoxifier; deobstruent for the liver, beneficial in the liver and joint diseases of the liver and kidney, remove excess heat from the liver and blood	Talaye Sabz Tuba Co./Iran
6	Ma’aljobon	Freeze-dried cheese water (whey)	Powder	Three full tablespoons; solve in warm water and take semi-warmly, 1–2 times a day	Liver and body detoxifier, black bile (sauda) repellent, blood purifier, laxative, exhilarant	Niak Pharmaceutical Co./Iran
7	Sauda Monzij	Jujube, licorice, fennel, lavender, melissa, fumitory, Iranian borage, maidenhair, Assyrian plum	Powder	Decoction/1–2 times a day	Sauda concoctive, resolvent of condensed matter, exhilarant, hypnotic	Talaye Sabz Tuba Co./ Iran
8	Itrifel Kishneezi	Coriander, Indian gooseberry, belleric myrobalan, myrobalan	Capsule	1–3 times a day with water	Strengthen the lower esophageal cardiac sphincter to prevent the release of stomach gas into the esophagus and the brain	Talaye Sabz Tuba Co./Iran
9	Dava-ye-Bagham	Thyme, nigella, mastic, ajwain	Capsule	One to two capsules a day with water	Relieves acidity and flatulence in the stomach, laxity, and lethargy of the body; repellent of phlegm in the organs, especially the stomach and liver (by dissolving and sweating)	Talaye Sabz Tuba Co./Iran
10	Hab al-Rahat	Aloe, senna, Chinese rhubarb, myrobalan	Tab	—One to two tabs every night	Relieve constipation	Niak Pharmaceutical Co./Iran
11	Moghl-e-Molayen	Aloe, mastic, myrobalan, Damask rose, guggul, leek	Capsule	One to two capsules every night with warm water	Relieve constipation	Talaye Sabz Tuba Co./Iran
12	Booqnaq	Eryngo	Distilled water	One cup/1–3 times a day	Reduce blood glucose level	Avand Co./Iran
13	Kasni	Chicory	Distilled water	Half of one glass/1–3 times a day	Remove excess heat from the liver and blood	Avand Co./Iran

Based on the pharmaceutical leaflet inside the drug packaging.

Statistical model

In longitudinal data analysis, two aspects of the data require modeling: (1) the mean response over time to characterize the change in response over time and the factors that influence this change and (2) within-individual covariance to capture the correlation between repeated measurements.

In this section, we utilize generalized least square (GLS) models to analyze blood glucose level (BGL) changes related to herbal medicine consumption, taking into account the longitudinal nature of the data. This approach allowed us to consider within-subject correlation by directly modeling the correlation structure. The GLS method was chosen over mixed-effects models due to its flexibility in creating a parametric model for the average response over time while addressing within-subject correlations.⁹ In addition, the GLS model can accommodate variations in follow-up times for different subjects. The GLS technique is used for inferential modeling with various longitudinal datasets as demonstrated in studies.^10
–12 We assume that a sample of $N$ subjects is measured repeatedly over time. Either by design or happenstance, the subjects may not be measured at the same time. We assume that there are n repeated measurements of the interesting outcome $Y_{i j}$ on the ith subject at the $j$ th visit time $t_{i j}$ . We consider a general linear regression model for changes in the mean response over time as follows: $Y_{i j} = β_{1} X_{i j 1} + β_{2} X_{i j 2} + \dots + β_{p} X_{ijp} + e_{i j}; j = 1, \dots, n,$ where $X_{i j} = (X_{i j 1}, \dots, X_{ijp})$ is the vector of covariates. Information about the time of observation, demographic variables, medication use, and other predictors can be expressed through this vector of covariates. Also as it is known, a confounder variable is the one that relates to both the independent variable and outcome of interest. To be able to adjust the result for potential confounders in a linear model approach, one has to add them to the model as covariates and see how the regression coefficients are modified. In general, if they have significant effect, they should be considered in the fitted model.

Commonly to insert an intercept in the mean model $X_{i j 1} = 1$ is assumed. In addition, $β_{1}, \dots, β_{p}$ are the unknown regression coefficients. The $e_{i j}$ ’s are random errors, with mean zero, and represent deviations of the $Y_{i j}$ ’s from their means, $E (Y_{i j} | X_{i j}) = β_{1} X_{i j 1} + β_{2} X_{i j 2} + \dots + β_{p} X_{ijp} .$

The vector of random errors for the $i$ th individual, $e_{i} = (e_{i 1}, \dots, e_{i n})$ , is assumed to follow a multivariate normal distribution with a zero-mean vector and an $n \times n$ covariance matrix $Σ$ . Let ${t_{i 1}, \dots, t_{i n}}$ denote the observation times for the $i$ th individual and assume that the error variance is constant across all measurement occasions, say $σ^{2}$ .

Because measurement occasions are not equally spaced over time, an exponential covariance pattern is assumed for $Σ$ , where the correlation between two times $j$ and $k$ for individual $i$ is calculated based on the following formula: $Corr (Y_{i j}, Y_{ik}) = ρ^{| t_{i j} - t_{i k} |},$ for the unknown $ρ \geq 0$ . According to this covariance structure, the correlation between any pair of repeated measures exponentially decreases with time separation between them.

In this study, the auxiliary variables were entered into the model in two scales. In the first case, if an herbal medicine was prescribed to the patient, the auxiliary variable for the patient was set to one and zero otherwise. This means that individual herbal medicines were considered as binary variables. In the second modeling framework, auxiliary variables were recorded based on the number of times medicine was taken per day. Therefore, if a medicine was not prescribed to an individual, a value of zero was considered.

To determine the best mean response model, we fitted several GLS models with different sets of auxiliary variables and used the Akaike information criterion (AIC) for model selection. AIC provides a measure of the information lost when using a model to represent the “true” relationship within the data. The model with the lowest AIC score is considered the most accurate representation of the true relationship. The models were estimated using maximum likelihood method with the final models estimated using the restricted maximum likelihood model.

R software version 4.2.1 was used to conduct our analysis by utilizing the gls function found in the nlme package developed by Pinheiro & Bates¹³ for generating GLS models. The stepAIC function was used to select the best model.

Results

In this section, the results of the exploratory data analysis are presented. This includes describing the demographic features of the study participants at baseline and the end of two subsequent visits and changes in FBS. In addition, the results of the fitted marginal model (explained in the “Model fitting” Section) and the interpretation of the significant parameters are reported here.

Descriptive statistics

Table 2 shows the summary statistics of the variables measured for all participants at the time of the initial visit. These variables include the patient’s demographic characteristics, disease history, and status. Based on the findings reported in Table 2, among 47 patients at the initial visit, more than half were women, and the mean age of the study participants was about 53 years. The mean weight of the participants was 81.78 kg, and the mean duration of T2DM was 9.49 years. The FBS of these individuals ranged from 99 to 375 mg/dL, and the BS ranged from 92 to 485 mg/dL.

Table 2.

Demographic Characteristics of All Study Participants at Baseline

Categorical variables (n = sample size)	Frequency	Percent
Gender (n = 47)
Male	21	44.7
Female	26	55.3

Continuous variables (n = sample size)	Unit of measurement	Minimum	Maximum	Mean	Standard deviation
Age (n = 45)	Years	30	75	53.26	11.57
Height (n = 36)	Centimeters	148	193	165.52	9.91
Weight (n = 42)	Kilograms	56	173	81.78	20.12
Duration of diabetes (n = 40)	Years	0.08	30	9.49	8.00
FBS (n = 43)	mg/dL	99	375	206.79	82.59
BS (n = 28)	mg/dL	92	485	255.42	100.88

BS, blood sugar; FBS, fasting blood sugar.

Only 27 participants attended two follow-up visits. Twenty patients left the study after the initial visit for various reasons, including the COVID-19 pandemic, and the high cost of herbal medicines that are not covered by insurance. The baseline summary statistics for the subset of complete cases over the two follow-up visits are presented in Table 3.

Table 3.

Demographic Characteristics at Baseline for Complete Cases during the Follow-up Period

Categorical variables (n = 27)	Frequency	Percent
Gender
Male	14	48.1
Female	13	51.9

Continuous variables (n = 27)	Minimum	Maximum	Mean	Standard deviation
Age	30	75	51.62	12.80
Height	150	193	167.55	10.83
Weight	58	113	80.29	15.03
Duration of diabetes	0.08	30	7.86	8.20
FBS	112	375	219.04	84.23
BS	92	450	254.13	104.62

According to this table, more balance has been established between the two gender groups for the set of complete cases. The average age, weight, height, and FBS are almost similar to Table 2, and the only notable difference is in baseline FBS, which is higher for complete cases compared with the whole sample.

Figure 2 shows the frequency of study participants in terms of herbal medicines used to reduce high BGL at the two follow-up times. At the time of the initial visit, the “diabetes powder” was the most prescribed medication (81.5%), whereas the “Moghl-e-Molayen” capsule was the least prescribed medication (11.1%). In the second follow-up visit (after receiving the baseline prescription), again “diabetes powder” was the most prescribed medication for the treatment of high blood glucose (92.6%). The medications that had the lowest number of prescriptions were the “Habb al-Rahat” capsule, “Moghl-e-Molayen” capsule, and “Itrifel Kishneezi” capsule (7.4% each) (Table 4).

FIG. 2.

Bar plots of prescription frequencies of herbal medicines at two follow-up visits.

Table 4.

Frequency of Study Participants According to the Herbal Medicines Prescribed at the First and Second Visit

	Visit time
	First follow-up visit (n = 27)		Second follow-up visit (n = 27)
Herbal medicine	Frequency of prescription	p-value	Frequency of prescription	p-value
Diabetes Powder	22 (81.5%)	0.038	25 (92.6%)	0.467
Diabetes Capsule	13 (48.2%)	0.57	18 (66.7%)	0.402
Diabetes Oil	10 (37.0%)	0.515	8 (29.6%)	0.975
Hab-e-Amber Momiai	5 (18.5%)	0.822	22 (81.5%)	0.102
Kabed Capsule	21 (77.8%)	0.816	16 (59.3%)	0.094
Ma’aljobon	8 (29.6%)	0.859	14 (51.9%)	0.235
Sauda Monzij Powder	10 (37.0%)	0.642	7 (26.0%)	0.545
Itrifel Kishneezi Capsule	3 (11.11%)	0.864	2 (7.4%)	0.406
Dava-ye-Bagham Capsule	5 (18.5%)	0.712	4 (14.8%)	0.529
Hab al-Rahat Capsule	3 (11.11%)	0.569	2 (7.4%)	0.136
Moghl-e-Molayen Capsule	2 (7.4%)	0.008	2 (7.4%)	0.196
Booqnaq	10 (37.0%)	0.933	10 (37.0%)	0.962
Kasni	5 (18.5%)	0.357	3 (11.11%)	0.292

The p-values are related to the t-test of the equality of the average FBS between users and nonusers of each medicine at each time.

The changes in the mean FBS of the study participants at the three measurement times are shown in Figure 3. It can be concluded that the mean FBS was considerably decreased over time, which is consistent with effectiveness of herbal medicines in reducing the average FBS of the observed sample.

FIG. 3.

The change in mean FBS of study participants across the two follow-up visits. FBS, fasting blood sugar.

Also, to visualize the impact of different herbal medicines’ use on the FBS changes, Figure 4 is given. In this figure, which includes line graphs for different herbal medicines, the two groups of users and non-users are shown by solid and dashed lines, respectively. According to these graphs, using Diabetes Capsule led to lower FBS at the final visit. Although Itrifel Kishneezi Capsule seems to have lower average FBS through time for their users compared with nonusers, the slope of the user’s line is very low (nearly 0). This is due to the fact that it has been used by those with lower baseline FBS compared with nonusers.

FIG. 4.

Line graphs of FBS changes over time for different herbal medicines.

FIG. 4.

(Continued).

FIG. 4.

(Continued).

Model fitting

Two general linear models with exponential covariance patterns were fitted to the above-described longitudinal cohort data. In the first model (Model I), to assess the effect of prescribing each herbal medicine to compare users and nonusers, binary predictors were included in the model to indicate whether each drug was prescribed. In the second model (Model II), continuous predictors showing the dose of medicine prescribed for each patient were included to analyze the effect of herbal medicine dosages on the course of diabetes. Since the baseline FBSs are nonhomogeneous baseline FBS’s, we have applied the model to those individuals with a baseline FBS of less than 327 mmol/L (23 patients). The initial comprehensive model included all potential factors affecting BGL. These factors include herbal medicine prescription status (specified in Table 1) as well as additional variables related to the individual’s characteristics and disease status (specified in Table 3). To search for the best model with the optimal set of significant predictors for the FBS response variable, stepwise model selection was performed using the stepAIC function available in the R software assuming 0.2 significance level. The parameter estimations of the two stepwise selected models are presented in Tables 5 and 6.

Table 5.

Results of Parameter Estimation of Model I: A Longitudinal General Linear Model with an Exponential Covariance Pattern for the Outcome of FBS, Including Binary Indicators of Herbal Medicine Use

Coefficients	Estimate	Std. Error	t-Value	^a p-Value
Intercept	136.23	14.86	9.17	0.000
Baseline FBS	0.29	0.06	4.72	0.000
Sex (Male)	−26.46	8.83	−2.99	0.005
Diabetes Capsule	−17.14	8.34	−2.06	0.046
Hab-e-Amber Momiai	−20.43	11.88	−1.72	0.093
Hab al-Rahat Capsule	−19.12	13.58	−1.41	0.167

p < 0.05 was considered as significant level in this study.

Table 6.

Results of Parameter Estimation of Model II: A Longitudinal General Linear Model with an Exponential Covariance Pattern for the Outcome of BGL, Including Herbal Medicine Dosages as Continuous Variables

Coefficients	Estimate	Std. Error	t-Value	^a p-Value
Intercept	132.30	14.17	9.34	0.000
Baseline FBS	0.33	0.06	5.25	0.000
Sex (Male)	−28.22	8.53	−3.31	0.002
Diabetes Capsule	−15.22	6.00	6.00	0.015
Hab-e-Amber Momiai	−14.14	6.89	−2.05	0.047
Hab al-Rahat Capsule	−9.80	6.64	−1.48	0.148

p < 0.05 was considered as significant level in this study.

BGL, blood glucose level.

The results shown in Table 5 (Model I) indicate that the baseline FBS has a significant relationship with BGL at the follow-up visits (p = 0.000). Specifically, controlling for other variables, each unit increase in the baseline FBS, leads to a higher average BGL of 0.29 mmol/L in the next visits. Also, the mean FBS at follow-up for men is about 26.46 mmol/L lower compared with females, assuming the same baseline FBS and the status of taking or not taking Diabetes Capsule, Hab-e-Amber Momiai (Habb-e-Amber Momyaie/Hab Ambar Momiai/Habbe/Amber Momyaee), and Hab al-Rahat Capsule for both gender groups. Furthermore, controlling for other variables, the use of the “Diabetes Capsule” significantly reduced the average FBS by 17.14 mmol/L compared with non-use (p = 0.046). On the contrary, controlling for other variables, there is some evidence that the use of “Hab-e-Amber Momiai” reduced individuals’ average FBS by 20.43 mmol/L compared with nonuse (p = 0.093). Also, the model shows some slight evidence that the use of “Hab al-Rahat Capsule” reduced the average FBS by 19.12 mmol/L compared with non-use (p = 0.167). The statistical significance of the findings for the last two drugs may be increased by increasing the sample size.

The results shown in Table 6 (Model II) also indicate that, controlling for other variables, patients with higher baseline FBS had a mean FBS of 0.33 mmol/L higher at follow-up visits (p = 0.000). On the other hand, controlling for other variables, males have about 28.22 unit lower average FBS compared to females (p = 0.002). Also, for each unit increase in the consumption dosage of “Diabetes Capsules,” the average FBS decreased significantly by 15.22 mmol/L (p = 0.015). In addition, controlling for other variables, a significant decrease of 14.14 mmol/L in the average FBS of individuals was observed for each unit increase in the consumption dosage of “Hab-e-Amber Momiai” (p = 0.047). Another finding was that, controlling for other variables, for each unit increase in the consumption dosage of “Hab al-Rahat Capsule “the average FBS of patients had a less significant reduction of 9.8 mmol/L (p = 0.148).

Discussion

This study was designed to test the usefulness of applying general linear models with exponential covariance structure to a prospective observational cohort of patients with T2DM treated with individualized herbal medicines derived from PM. While the study was unable to reach its anticipated sample size due to the confluence of the COVID-19 pandemic and a sudden escalation in the cost of medicinal plants in Iran, the models identified the herbal medicines associated with the greatest patient improvement, and the findings remain compelling and noteworthy, despite these constraints.

Like many complementary therapies, PM takes a personalized yet holistic approach.¹⁴ Holistic care focusing on physical, emotional, social, cultural, and spiritual well-being has proven effective in managing DM.¹⁵

The eminent Persian scholar Avicenna (980–1073 AD), in his invaluable medical encyclopedia Canon of Medicine, classifies diseases into two main categories: elemental and participatory. According to this classification, an elemental disease directly affects the core functioning of an organ, while a participatory disease arises in one organ due to an underlying pathology in another. Avicenna posits that organs can influence each other’s health because of their physical proximity, their symbiotic roles as servants and masters, their origination of vital faculties, or their place along pathways for waste elimination. Thus, in Avicenna’s view, a symptom manifesting in one organ may stem from dysfunctions across multiple, interconnected organs.^16,17 The PM specialist always considers these possibilities by evaluating participatory illnesses alongside the chief complaint when diagnosing and treating patients. Based on this premise, the study began without any preconceived assumptions by the physician about the number and type of prescribed drugs. The antidiabetic medications specified by the pharmaceutical companies in the drugs’ leaflets (Table 1) were prescribed for patients, and there were variations in the type, dosing, and timing. The researcher, solely as an observer without intervening, documented the individualized prescriptions for each diabetic patient. After three visits for all participants, ∼13 frequently prescribed drugs emerged, while infrequently prescribed ones were excluded (Table 1). Given the diversity of herbal medicines prescribed to diabetic patients and the nonuniformity of patient visit intervals, a general linear model with an exponential covariance pattern was used.^7,8

The results of the fitted models indicated a significant effect of baseline FBS on the following BGL at the two upcoming visits. As expected, subjects with higher FBS levels at baseline would experience slightly higher BGL at the second and third visits. This is not unexpected and simply shows that the model functions as anticipated.

The study demonstrated that gender influences changes in BGL; on average, males had 28.22 units lower FBS compared with females (p = 0.002). This significant difference indicates gender should be controlled for as a blocking factor in similar studies in T2DM. A study found that prediabetes, diabetes, awareness, and BGL control were higher in females, yet females had lower treatment rates. Though diabetes awareness grew in both genders, the improvement was greater in men (30.9%) than women (26.5%).¹⁸ The difference in treatment response observed in males in our study may be due to greater diabetes awareness in men, as reported in this study.

The results of the fitted models showed that when holding other drugs, baseline FBS, and gender constant, “Diabetes Capsule,” among all prescribed drugs, drop the FBS by 17.14 mmol/L (p = 0.046), and it reduces FBS by 15.22 mmol/L per unit increase (up to 2 tablets/day, p = 0.015). “Hab-e-Amber Momiai” was also linked to a 14.14 mmol/L FBS drop per unit increase (up to 2 tablets/day, p = 0.047).

“Diabetes Capsule” is a compound herbal product specifically marketed for diabetes treatment (Table 1), and its antidiabetic properties are attributed to its key ingredient, Citrullus colocynthis (bitter apple or Hanzal), which has long been known for its hypoglycemic effects all over the world.^19,20 C. colocynthis is an important source of phytochemicals that exhibit potential antioxidant, amylase, and glucosidase inhibitory activities. Oxidative stress is understood to play a major role in the pathogenesis of diabetic complications. The enzyme α-amylase facilitates starch breakdown in the gastrointestinal tract, leading to glucose release and absorption into circulation. Amylase inhibition can reduce starch digestion, thereby decreasing postprandial glucose levels. α-glucosidase is involved in cleaving oligosaccharides and glycosides into monosaccharides such as glucose. Inhibition of α-glucosidase activity can delay glucose release from carbohydrates and lower postprandial glucose levels in diabetic patients.²⁰ C. colocynthis median lethal dose (LD50) is 200 mg/kg, causing histological changes in the small intestine, liver, and kidney.¹⁹ Each 250 mg “Diabetes Capsule” contains 75 mg of C. colocynthis along with three other ingredients—tragacanth and Arabic gum as correctives and anise (Pimpinella anisum ((aniseed)—to enhance its antidiabetic effect. In vitro studies on the methanolic extract and ethyl acetate fraction of aniseed extracts have demonstrated significant free radical scavenging capacities (2,2′-azino-bis(3-ethylbenzothiazoline-6-sulfonic acid); ABTS) (2,2-diphenyl-1-picrylhydrazyl; DPPH assays) along with antidiabetic and hypolipidemic effects. These are achieved through inhibition of α-amylase, α-glucosidase, β-Hydroxy β-methylglutaryl-CoA (HMG-CoA) reductase, and pancreatic lipase activities. Furthermore, in vivo studies supplementing patients with diabetes with 5 g/day of aniseeds for 60 days provided evidence for antihyperglycemic, hypolipidemic, and antioxidant properties.²¹

On the contrary, the prescription of Hab-e-Amber Momiai, aimed at enhancing sexual potency and strengthening the body (Table 1), showed significant results in lowering BGL with an increase in dosage (p = 0.047). According to the claims made by the Hamdard Laboratories that manufactures it, this drug acts as both a physical and nervous system tonic. It enhances masculine vigor, improves the function of reproductive organs, and increases energy levels throughout the genital system.²² This unpredicted result led us to focus on the antidiabetic properties of the individual ingredients in this compound pill. We found that many components of this pill have demonstrated hypoglycemic effects in previous animal or human studies. The statistically meaningful reduction in BGL by this multi-herb formulation has been substantiated by the antidiabetic characteristics of its constituents.

The pill is made from various plants such as cinnamon (Darchin) (Cinnamomum verum),^23

–26 pistachio (Peste) (Pistacia vera),^27,28 clove (Mikhak) (Cizigium aromaticum),^29,30 Egyptian asparagus (Shaghaghol) (Asparagus racemosus),^31,32 bamboo (Tabashir) (Bambusa arundinacea),^33,34 nutmeg (Joze boa) (Myristica fragrans),^35,36 white turmeric (Jadvar) (Curcuma zedoaria),^37,38 ginger (Zangebil) (Zingiber officinale),^24,28,39,40 peony (Uod-el Salib) (Paeonia spp.),^41,42 mastic (Mastaki) (Pistacia lentiscus),^43
–45 amber (Anbar Ashab) (Ambergris),^46,47 mummy, and a few other ingredients. According to the mentioned studies, the antidiabetic effects of these plants are likely attributed to a multifaceted mechanism involving the inhibition of alpha-amylase and alpha-glucosidase, thereby hindering carbohydrate digestion and absorption. In addition, these plants may enhance insulin secretion and function, reduce oxidative stress and improve antioxidant activities in pancreatic tissues, increase insulin sensitivity/synthesis, protect ß cells in pancreatic islets, lower fat accumulation, and augment glucose uptake by tissues, collectively leading to a significant reduction in blood glucose levels.

Based on the proven effects of many of its components on BGL, it appears that this drug may have helped improve the cellular insulin response and reduce FBS by strengthening the main organs (heart, brain, and liver) and increasing the instinctive heat and regulatory power of the body. Therefore, this compound pill, originally formulated for other purposes and prescribed as an ancillary medicine, could potentially become a primary antidiabetic agent recommended for all diabetic patients regardless of sexual dysfunction, of course, if there was no other contraindication for its prescription.

This study’s findings underscore the necessity for preliminary observational research across all complementary medical systems that employ individualized, multimedication prescriptions. Such an approach aligns with the fundamental principles of holistic and personalized medicine, which inherently diverge from homogeneous treatment protocols.⁵ For holistic medicine practitioners, it is crucial to determine the optimal drugs, dosages, and combinations that enhance therapeutic efficacy for each patient. In this regard, the proposed statistical model presents a valuable tool that can assist them while adhering to TM principles. Furthermore, these findings have significant practical implications for clinicians who integrate complementary and alternative medicine approaches into diabetes management strategies. The study highlights two herbal formulations, “Diabetes Capsule” and “Hab-e-Amber Momiai,” as promising candidates for further investigation in the management of T2DM. While their specific mechanisms of action require additional research, the observed effectiveness in this study suggests their potential as complementary therapies for T2DM management.

Sufficiently large studies of this type can reveal which body systems typically require intervention for each disease. For example, this study’s 13 variables suggest diabetic patients generally need supportive treatments such as gastrointestinal regulation (Itrifel Kishneezi and Dava-ye-Bagham), constipation relief (Hab al-Rahat and Moghl-e-Molayen), anxiety and stress reduction (Sauda Monzij), and liver detoxification (Kabed Capsule, Kasni distilled water, and Ma’aljobon), explaining the physician’s varied formulations to address possible influences of participating organs (Table 1).

These low-cost preliminary studies can also lead the researcher to find new uses for previously known drugs, as occurred in this study with regard to Hab-e-Amber Momiai.

It is important to note that this type of study, like other observational studies, cannot necessarily prove causality. It is advisable that future studies with a larger sample size also gather information on confounding factors that could impact BGL, such as the hypoglycemic chemical drugs used and nonpharmacological interventions such as diet and intense physical activity. If these factors significantly influenced the measured outcome and truly acted as confounders, they should be incorporated into the model and their impact adjusted for. A model with a larger sample size in which all likely confounding factors are controlled for may strengthen the case for causality. Unfortunately, this approach was not feasible in the present study, which aimed to introduce a statistical method and ended with a small sample size due to the COVID-19 pandemic. In a study with a small sample size, this type of approach can only produce hypotheses for conducting subsequent clinical trials.

As these initial observational studies can provide ideas for subsequent clinical trials, it is recommended that the effects of “Diabetes Capsule” and “Hab-e-Amber Momiai” versus placebo be examined through a standardized, homogeneous clinical trial in the future.

Conclusion

This preliminary study aimed to propose a suitable methodology for personalized medical studies and validate its applicability through a small-scale study involving patients with T2DM.

Unlike controlled clinical trials, the participants were administered herbal remedies tailored to their unique attributes as determined by a personalized medicine approach of PM. Due to the limited sample size, some linear models were fitted. Also, as the number of features was slightly large relative to the sample size, subset-selection approaches such as step-wise regression were used to reduce feature space and obtained some smaller models that are hoped to be among the best ones.

Further evaluation of the proposed model in future research within the realm of traditional medicine is recommended, utilizing a larger sample size. To enhance result accuracy and personalize the outcomes to a greater extent, a good strategy would categorize the obtained data according to the constitutional diagnoses of traditional medicine, classifying patients into more similar groups, such as four temperamental groups⁴⁸ based on PM principles, Unani and Anthroposophic Medicine, or different patterns of syndrome based on Traditional Chinese Medicine⁴⁹ or three Doshas in Ayurveda⁵⁰ could be beneficial.⁵¹ In the context of PM, employing a standard temperament questionnaire tool during the study can facilitate the analysis and reporting of results based on temperamental groups after the study and during the analysis conducted by a statistician. This approach will lead to more precise and personalized results.

Having a richer dataset, as future work, we are interested in developing statistical and machine learning-based models to evaluate and predict personalized treatments for each patient according to their classical and traditional medicine characteristics. These can include more complex nonlinear models such as polynomial or spline-based models and decision trees to identify potential interactions automatically.

It is important to note that the introduced methodology is not new for statisticians, but it is innovative in terms of its application in traditional medicine studies.

Conducting this type of preliminary studies that involve assessing the impact of multiple prescribed medications concurrently for a particular health issue based on each patient’s attributes reduces the intricacies of holistic medicine and personalized care. Without this initial investigation, researchers would need to perform numerous trials on each prescribed medication. However, by conducting this initial assessment and pinpointing the most promising medication(s), researchers only focus on them in subsequent clinical trials, streamlining the process, and eliminating less effective medications from further consideration and extensive research scrutiny. Although these studies may not definitively establish causal relationships, they provide significant support for forthcoming clinical trials.

This proposed methodology represents a significant step forward in enhancing patient care within the holistic medicine sphere. Healthcare professionals in the person-centered traditional medicine benefit greatly from this analytical tool as it can potentially optimize patient outcomes and elevate the quality of care provided. Throughout the study, no participants reported severe adverse effects necessitating medical intervention.

Confirmation statement

Each author listed in this article affirms that the research presented is supported by an institution primarily engaged in educational or research activities.

Ethical Approval

The study was conducted in accordance with the Declaration of Helsinki and approved by the research ethics committee of AJA University of Medical Sciences (IR.AJAUMS.REC.1400.010) and is signed by director and secretary of university/regional research ethics committee of AJA University of Medical Sciences.

Footnotes

Acknowledgments

The authors would like to express their gratitude to Dr. Maryam Taghavi Shirazi for her guidance and valuable comments that significantly enriched this article. The authors also thank all the patients who participated in this study.

Authors’ Contributions

This study was designed by A.Z. and R.Gh. Data collection was performed by R.Gh. and R.Kh. Data analysis and modeling were done by S.E.M. and R.Kh. R.Gh., S.EM., A.S., L.S.W., and R.Gh. prepared the draft version, and all authors critically revised it. All authors read and approved the final version.

Data Availability Statement

All data generated or analyzed during this study are included in this article. Further inquiries can be directed to the corresponding author.

Author Disclosure Statement

Nothing to declare.

Funding Information

This study was funded by the AJA University of Medical Sciences with grant number 97001611.

References

Naghizadeh

, Salamat

, Hamzeian

, et al. IrGO: Iranian traditional medicine General Ontology and knowledge base. J Biomed Semant, 2021; 12(1):9; doi: 10.1186/s13326-021-00237-1

Tafazoli

, Tavakoli

, Mosaffa-Jahromi

, et al. Approach of Persian medicine to health and disease. Adv Integr Med, 2022; 9(1):3–8; doi: 10.1016/j.aimed.2021.07.007

Mohammadi Kenari

, Taqavi Shirazi

, Jafari

, et al. Study of patient satisfaction with Persian medicine in referrals to Behesht Persian Medicine Clinic of Iran University of Medical Sciences in 2016. Altern Integr Med, 2018; 07(01):257; doi: 10.4172/2327-5162.1000257

WHO Traditional Medicine Advisory Team. WHO Traditional Medicine Global Summit 2023 meeting report: Gujarat Declaration. Available from: https://www.who.int/publications/m/item/who-traditional-medicine-summit-2023-meeting-report–gujarat-declaration [Last accessed: January 26, 2024 ].

Chen

, Wang

, Ding

, et al. Problems lowering the study quality in traditional medicine, introspection from an example of meta-analysis of acupuncture. BMC Complement Med Ther, 2020; 20(1):41; doi: 10.1186/s12906-019-2806-z

Ray

, Fitzpatrick

, Golubic

, et al. Oxford handbook of clinical and healthcare research. Oxford University Press: 2016.

Fitzmaurice

, Laird

, Ware

. Applied Longitudinal Analysis. 2 ed.; Wiley: United States of America, 2011.

Caruana

, Roman

, Hernández-Sánchez

, et al. Longitudinal studies. J Thorac Dis, 2015; 7(11):E537–40; doi: 10.3978/j.issn.2072-1439.2015.10.63

Harrell

. Regression Modeling Strategies. 2 ed.; Springer New York, 2015.

10.

Askim

, Bernhardt

, Churilov

, et al. Changes in physical activity and related functional and disability levels in the first six months after stroke: A longitudinal follow-up study. J Rehabil Med, 2013; 45(5):423–428; doi: 10.2340/16501977-1137

11.

Fabbri

, An

, Gonzalez-Freire

, et al. Bioavailable testosterone linearly declines over a wide age spectrum in men and women from the Baltimore Longitudinal Study of Aging. J Gerontol A Biol Sci Med Sci, 2016; 71(9):1202–1209; doi: 10.1093/gerona/glw021

12.

Lindblad

, Nelson

, Zeiler

, et al. Influence of blood-brain barrier integrity on brain protein biomarker clearance in severe traumatic brain injury: A longitudinal prospective study. J Neurotrauma, 2020; 37(12):1381–1391; doi: 10.1089/neu.2019.6741

13.

Pinheiro

, Bates

, Saikat

, et al. nlme: Linear and Nonlinear Mixed Effects Models. Available from: https://CRAN.R-project.org/package=nlme

14.

Yuan

. Toward holistic medicine and holistic biology: Life sciences after precision medicine and systems biology. Front Life Sci, 2019; 12(1):14–26; doi: 10.1080/21553769.2019.1682064

15.

Juanamasta

, Aungsuroch

, Gunawan

, et al. Holistic care management of diabetes mellitus: An integrative review. Int J Prev Med, 2021; 12:69; doi: 10.4103/ijpvm.IJPVM_402_20

16.

Avicenna. Al Qanun Fil Tibb (Canon of Medicine). Jamia Hamdard Printing Press: New Delhi, 1988; Vol. 2.

17.

Yousofpour

, Kamalinejad

, Esfahani

, et al. Role of heart and its diseases in the etiology of depression according to Avicenna’s point of view and its comparison with views of classic medicine. Int J Prev Med, 2015; 6:49; doi: 10.4103/2008-7802.158178

18.

Khodakarami

, Abdi

, Ahmadnezhad

, et al. Prevalence, awareness, treatment and control of diabetes among Iranian population: Results of four national cross-sectional STEPwise approach to surveillance surveys. BMC Public Health, 2022; 22(1):1216; doi: 10.1186/s12889-022-13627-6

19.

Kamil

, Ahmad

, Abdallah

. Scientific studies on Citrullus colocynthis with special reference to anti-hyperglycemic activity. J Pharm Pharmacol, 2020; doi: 10.17265/2328-2150/2020.09.004

20.

Parveen

, Khan

, Zahiruddin

, et al. TLC-bioautographic evaluation for high-throughput screening and identification of free radical scavenging and antidiabetic compounds from traditional unani medicinal plant: Citrullus colocynthis Schrad. J AOAC Int, 2020; 103(3):669–677; doi: 10.5740/jaoacint.19-0287

21.

Shobha

, Andallu

. Antioxidant, anti-diabetic and hypolipidemic effects of aniseeds (Pimpinella anisum L.): in vitro and in vivo studies. Jcmah, 2018; 5(2); doi: 10.19080/JCMAH.2018.05.555656

22.

Fazli

, Faheem

. The effect of Hab Ambar Momiai on male reproductive physiology. Hamdard Medicus, 1996.

23.

Al Dhaheri

, Alkhatib

, Feehan

, et al. The effect of therapeutic doses of culinary spices in metabolic syndrome: A randomized controlled trial. Nutrients, 2024; 16(11); doi: 10.3390/nu16111685

24.

Zelicha

, Yang

, Henning

, et al. Effect of cinnamon spice on continuously monitored glycemic response in adults with prediabetes: A 4-week randomized controlled crossover trial. Am J Clin Nutr, 2024; 119(3):649–657; doi: 10.1016/j.ajcnut.2024.01.008

25.

Shen

, Fukushima

, Ito

, et al. Verification of the antidiabetic effects of cinnamon (Cinnamomum zeylanicum) using insulin-uncontrolled type 1 diabetic rats and cultured adipocytes. Biosci Biotechnol Biochem, 2010; 74(12):2418–2425; doi: 10.1271/bbb.100453

26.

Mandal

, Sharma

, Yadav

SRM

, et al. Efficacy of young Cinnamomum zeylanicum blume bark on hyperglycemia and PTPase activity in type 2 diabetes. Cureus, 2023; 15(2):e35023; doi: 10.7759/cureus.35023

27.

Ribeiro

PVM

, Silva

, Almeida

, et al. Effect of chronic consumption of pistachios (Pistacia vera L.) on glucose metabolism in pre-diabetics and type 2 diabetics: A systematic review. Crit Rev Food Sci Nutr, 2019; 59(7):1115–1123; doi: 10.1080/10408398.2017.1392290

28.

Carvalho

GCN

, Lira-Neto

JCG

, Araújo

MFMd

, et al. Effectiveness of ginger in reducing metabolic levels in people with diabetes: A randomized clinical trial. Rev Lat Am Enfermagem, 2020; 28:e3369; doi: 10.1590/1518-8345.3870.3369

29.

Kuroda

, Mimaki

, Ohtomo

, et al. Hypoglycemic effects of clove (Syzygium aromaticum flower buds) on genetically diabetic KK-A y mice and identification of the active ingredients. J Nat Med, 2012; 66(2):394–399; doi: 10.1007/s11418-011-0593-z

30.

Mohan

, Jose

, Mulakkal

, et al. Water-soluble polyphenol-rich clove extract lowers pre-and post-prandial blood glucose levels in healthy and prediabetic volunteers: An open label pilot study. BMC Complement Altern Med, 2019; 19(1):99–99; doi: 10.1186/s12906-019-2507-7

31.

Vadivelan

, Krishnan

, Kannan

. Antidiabetic potential of Asparagus racemosus Willd leaf extracts through inhibition of α-amylase and α-glucosidase. J Tradit Complement Med, 2019; 9(1):1–4; doi: 10.1016/j.jtcme.2017.10.004

32.

Hannan

, Ali

, Khaleque

, et al. Antihyperglycaemic activity of Asparagus racemosus roots is partly mediated by inhibition of carbohydrate digestion and absorption, and enhancement of cellular insulin action. Br J Nutr, 2012; 107(9):1316–1323; doi: 10.1017/S0007114511004284

33.

Menaria

. Antidiabetic activity of leaves extract of Bambusa arundinacea . Pharm Chem J, 2016; 3(2):197–200.

34.

Macharla

, Goli

, Santhosha

, et al. Antdiabetic activity of Bambusa arundinaceae stem extracts on alloxan induced diabetic rats. J Chem Bio Phy Sci, 2012; 2(2):832.

35.

Arulmozhi

, Kurian

, Veeranjaneyulu

, et al. Antidiabetic and antihyperlipidemic effects of Myristica fragrans in animal models. Pharm Biol, 2007; 45(1):64–68; doi: 10.1080/13880200601028339

36.

Pashapoor

, Mashhadyrafie

, Mortazavi

. Ameliorative effect of Myristica fragrans (nutmeg) extract on oxidative status and histology of pancreas in alloxan induced diabetic rats. Folia Morphol (Warsz), 2020; 79(1):113–119; doi: 10.5603/FM.a2019.0052

37.

Rahmatullah

, Azam

MNK

, Pramanik

, et al. Antihyperglycemic activity evaluation of rhizomes of Curcuma zedoaria Christm. roscoe and fruits of Sonneratia caseolaris . Int J PharmTech Res, 2012; 4(1):125–129.

38.

Sadiq

, Shiv

, Vigneshwaran

, et al. A study on antidiabetic potency of mixture of powders of dried fruits of Eucalyptus globules and rhizomes of Curcuma zedoaria . Rjpbcs, 2011; 2(3):326–332.

39.

Akash

MSH

, Rehman

, Tariq

, et al. Zingiber officinale and type 2 diabetes mellitus: Evidence from experimental studies. Crit Rev Eukaryot Gene Expr, 2015; 25(2):91–112; doi: 10.1615/CritRevEukaryotGeneExpr.2015013358

40.

Shidfar

, Rajab

, Rahideh

, et al. The effect of ginger (Zingiber officinale) on glycemic markers in patients with type 2 diabetes. J Complement Integr Med, 2015; 12(2):165–170; doi: 10.1515/jcim-2014-0021

41.

Liu

, Li

, Qian

, et al. Phytochemical profiles and the hypoglycemic effects of tree peony seed coats. Food Funct, 2021; 12(23):11777–11789; doi: 10.1039/D1FO02341A

42.

Zhang

C-C

, Geng

C-A

, Huang

X-Y

, et al. Antidiabetic stilbenes from peony seeds with PTP1B, α-glucosidase, and DPPIV inhibitory activities. J Agric Food Chem, 2019; 67(24):6765–6772; doi: 10.1021/acs.jafc.9b01193

43.

Mehenni

, Atmani-Kilani

, Dumarçay

, et al. Hepatoprotective and antidiabetic effects of Pistacia lentiscus leaf and fruit extracts. J Food Drug Anal, 2016; 24(3):653–669; doi: 10.1016/j.jfda.2016.03.002

44.

Cherbal

, Kebieche

, Yilmaz

, et al. Antidiabetic and hypolipidemic activities of Algerian Pistachia lentiscus L. leaves extract in alloxan-induced diabetic rats. S Afr J Bot, 2017; 108:157–162; doi: 10.1016/j.sajb.2016.10.024

45.

Rehman

MSU

, Kamran

, Ahmad

, et al. Anti-diabetic activity of crude Pistacia lentiscus in alloxan-induced diabetes in rats. Bangladesh J Pharmacol, 2015; 10(3):543–547; doi: 10.3329/bjp.v10i3.23225

46.

Singh

, Khan MI

Badruddeen

, et al. Potential benefits of ambergris beyond perfume. Jcmah, 2018; 7(1); doi: 10.19080/JCMAH.2018.07.555705

47.

Olimat

. A review on ambergris perspective and modern chemical composition and pharmacology. Acad J Med Plants, 2020; 8:96–101; doi: 10.15413/ajmp.2020.0125

48.

Akhtari

, Moeini

, Mojahedi

, et al. Assessment the studies on the concept of Mizaj (temperament) in Persian Medicine. J Complement Integr Med (JCIM), 2020; 17(3):20180122; doi: 10.1515/jcim-2018-0122

49.

Lian

, Ni

, Shen

, et al. International traditional Chinese medicine guideline for diagnostic and treatment principles of diabetes. Ann Palliat Med, 2020; 9(4):2237–2250; doi: 10.21037/apm-19-271

50.

Thottapillil

, Kouser

, Kukkupuni

, et al. An ‘Ayurveda-Biology’ platform for integrative diabetes management. J Ethnopharmacol, 2021; 268:113575; doi: 10.1016/j.jep.2020.113575

51.

Baars

, Hamre

. Whole medical systems versus the system of conventional biomedicine: A critical, narrative review of similarities, differences, and factors that promote the integration process. Evid Based Complement Alternat Med, 2017; 2017:4904930; doi: 10.1155/2017/4904930