Fuzzy based approach for smart health monitoring systems using IoT devices

Abstract

IoT systems base devices are considered an excellent research domain owning to its expertise and applications in wide range of areas. IoT in health care domain is gaining attention due to its better access to the doctor and paramedical staff as well as sensor based studies which results in less man to man interacting and less fault in the data. The health care provider can easily access the vitals and various other medical parameters by even staying miles away from the patient. However, large amount of data transfer over various communication mediums results in more data traffic. This data transfer will require more power which will be utilized to transfer the data. To reduce this data traffic issues, an efficient method is used in this work in which only the data that is predominantly important to be send to the health care provider is send via the communication medium. Rule based fuzzy logic tool is used in this work for an elder patient having cardiac issues. Blood sugar (After eating), Blood pressues (systolic), Blood pressure (Diastolic) and cholesterol level are taken as the parameter that are examined for the patient and the medical treatment required is calculated. The rules are set on the basis of real time data and human knowledge. The results from the fuzzy logic interference shows that the health care provider will be alarmed using communication medium only when active or emergency medical treatment of the patient is required. A comparative study between the power utilized in normal data driven method and fuzzy method shows that the fuzzy method utilize 8 times less power than the normal method. The simulated and MAMDANI model calculated values shows less than 1% error which shows the accuracy of the work in health care domain.

Keywords

Fuzzy logic healthcare system Internet of Things

1 Introduction

The Internet of Things are considered as a world which can connect every kind of thing we can think of including accessing any information about nearly everything from anywhere in the world [1]. These smart objects and systems provide an ultimate path for linking change in surroundings to people anywhere in the world. IoT is creating and connecting any objects internally with the recent decade. Some advance applications of IoT includes, smart city [2], agricultural fields [3], smart climate [4] as well as smart parking [5] and smart health monitoring mechanism [6]. The most incredible application of use IoT is classified in the field of healthcare management which provides health and environment condition tracking facilities [7, 8].

In healthcare domain, Internet of Things plays vital role in various application including tacking of data about patients including measurement of vital parameters, physical, physiological, psychological and behavioral state. The data is collected automatically which results in less human error. The healthcare provider will get the data of the patient, with little or negligible error. Based on this in house data, the healthcare provider can easily classify the medical condition of the patient. The placement of IoT devices will results in an enhancement in the quality of diagnosis as well as false data collection due to human intervention [9, 10].

With advancement in the field of health monitoring, various IoT based systems and devices are used for monitoring of patients health parameters with high accuracy and stability [11, 12]. IoT devices provide a communication linkage between real time monitoring of vital signs of patients to the health care provider anywhere in the world. A patient with health issues in their elderly ages requires extensive care and monitoring to avoid any hazardous situation which can let to serious illness [13]. Al-Adhab et al. demonstrate an elderly patient care system which provide a constant monitoring of patient based on it condition, explicit and communication with its healthcare provider. Among various other age groups patients, elderly patients required constant monitoring to avoid any health related issue including Cardio Vascular diseases [14, 15]. This requires constant monitoring of the vital parameters including the cholesterol levels, blood sugar level and blood pressure (Both systolic and diastolic). A change in these parameters may lead to cardio vascular diseases like angina and stroke [16, 17].

The health care data is fetched from various sensors. The data is sent to the health-care provider using various communication channels. Huge traffic is generate when data from every sensor is transmitted using communication channels [18, 19]. This results in large amount of power used for the transmittion of the data. The elderly patients are required to be examimed multiple times in a day. The transmittion of the data will result in huge traffic when all the information while be send to the health-care provider. Fuzzy based data accusion system are widely used in such cases [20 –22].

In this work, an IoT fuzzy based event driven technique based real time data from the elderly patient is monitored based on the cholesterol levels, blood sugar level and blood pressure (Both systolic and diastolic) levels. On the basis of these parameters the output state of the patient in determined. If the patient need emergency treatment the command is directly sent to the healthcare provider for required treatment of the patient. Fuzzy based system provide a analysis of the data collect from the sensor and analyze the output state on the basis of the patient condition. This IoT based system can be used more efficiently without compromising with the quality of service of the system.

2 Fuzzy analysis

The fuzzy assisted technique may be employed in wide area of health applications. As a test case, we have tried to implement fuzzy assisted detection of elderly aged cardiac patient in a manner similar to the diagnositic process in real world.

Fuzzy analysis is considered an efficient method to analyze the data from the sensors and provide an output based on human thinking. The process as shown in Fig. 1 includes Imprecise and unreliable sensor readings which is send to fuzzifier consisting of fuzzy interference system. The rules in the fuzzy interference systems are set of the basis of knowledge based and real time data. The interference is then subjected to de-fuzzifier which provides the output data based on the input sensor data.

Fig. 1

Conceptual figure of fuzzy system.

Fuzzy works in the way human thought process works. In comparison to other algorithms based on probability theory, fuzzy logic is much more intuitive and easy to use.

The data collect from patient including the blood sugar level, blood pressure and cholesterol level is taken as input and the medical treatment required is taken as output in FIS as shown in Fig. 2.

Fig. 2

FIS of inputs and output.

Three membership functions are assigned to each input and ranges are assigned as shown in Fig. 3. For blood sugar level (After eating), the range is set as 170–300 mg/DL with membership functions of normal, impaired glucose and diabetics. For input blood pressure (Systolic), the range is set as 100–190 mmHg with membership functions of normal, hypertension, hypertensive crisis. For input blood pressure (diastolic), the range is set as 70–130 mmHg with membership functions of normal, hypertension, hypertensive crisis For input cholesterol level, the range is set as 100–300 mg/dL with membership functions of optimal, intermediate and high. The output membership function are set as normal monitoring, active monitoring and emergency monitoring as shown in Fig. 4.

Fig. 3

Membership functions for input (a)Blood Sugar (After Eating) (b) Blood Pressure (Systolic) (c) Blood Pressure (Diastolic) (d) Cholesterol Level.

Fig. 4

Membership functions of output medical treatment required.

Table 1 shows the membership functions for the input and output and their corresponding ranges. Rules are selected on the basis of the number of inputs.

Table 1

Input and output membership functions and ranges

Input/output	Membership function	Ranges
Blood Sugar (After Eating) (Mg/DL)	Normal	170–220
	Impaired Glucose	190–240
	Diabetic	230–300
Blood Pressure (Systolic) (mm/Hg)	Normal	100–135
	Hyper Tension	120–155
	Hypertensive Crisis	145–190
Blood Pressure (Diastolic) (mm/Hg)	Normal	70–95
	Hyper Tension	80–105
	Hypertensive Crisis	95–130
Cholesterol Level (mg/DL)	Optimal	100–200
	Intermediate	150–250
	High	200–300
Medical Treatment Required	Normal Monitoring	0–0.5
	Active Monitoring	0.3–0.7
	Emergency Monitoring	0.5–1

While using MAMDANI model, the number of rules are calculated by 3n, where n is the number of inputs. For 4 inputs, 81 rules are selected based on knowledgeable and human thinking based data.

The rules are selected in the way in which the crisis and critical situation of the patient lies in the output of emergency treatment. Few of the rules out of the 81 rules selected based on real life cases are stated below.

When Blood Sugar level is normal, blood pressure(systolic) is normal, blood pressure(diastolic) is normal and cholesterol level is normal, then output of normal monitoring is selected.

When Blood Sugar level is normal, blood pressure(systolic) is normal, blood pressure(diastolic) is hypertensive crisis and cholesterol level is high, then output of emergency monitoring is selected.

When Blood Sugar level is impaired glucose, blood pressure(systolic) is hypertension, blood pressure(diastolic) is hypertension and cholesterol level is intermediate, then output of active monitoring is selected.

When Blood Sugar level is diabetic, blood pressure(systolic) is hypertensive crisis, blood pressure(diastolic) is hypertensive crisis and cholesterol level is high, then output of emergency monitoring is selected.

When Blood Sugar level is diabetics, blood pressure(systolic) is normal, blood pressure(diastolic) is hypertension and cholesterol level is high, then output of active monitoring is selected.

When Blood Sugar level is normal, blood pressure(systolic) is normal, blood pressure(diastolic) is normal and cholesterol level is intermediate, then output of normal monitoring is selected.

3 Results and Discussion

Based on the rules, the 3D graphs of input and its effect on output is shown in Fig. 5. Figure 5 (a) shows a 3D graph of Blood pressure systolic and diastolic with respect to output medical treatrment required. Figure 5 (b) shows a 3D graph of Blood sugar (After Eating) and cholesterol level with respect to output medical treatrment required.

Fig. 5

3D graphs of output medical treatment required along (a) Blood pressure systolic and diastolic (b) Blood sugar (After Eating) and cholesterol level.

The 3D graphs in Fig. 5 (a) shows that an emergency treatment is required with systolic and diastolic blood pressure reaches hypertensive crisis level. However active monitoring is required in case of one of the systolic and diastolic blood pressure lies in the range of hypertension. Similarly blood sugar (after eating in the range of normal and impaired glucose require normal and active monitoring. Intermediate and high level of cholesterol require active monitoring.

The healthcare provider is given an emergency treatment output in a few cases. One of the case is when the patient vital sign are in in blood pressure (systolic) and blood pressure (diastolic) is in the range of hypertensive crisis and blood sugar (After eating) is diabetic and cholesterol level is high. This data is send to the health care provider using the communication channel. In short, only in those cases where patient required emergency treatment, as per the rules set in the fuzzy logic controller, the data is sent to the health-care provider.

A comparative study was performed to analyze the power utilized with and without using fuzzy logic controller while fetching data from the sensor and sending it to the health-care provider. The power consumption reduce 8 times while using event driven approach using fuzzy logic in place of direct data communication of sensor data to health-care provider.

Figure 6 shows a comparison between the power utilized when data is directly send to the health care provider and data is send after the event driven approach based on fuzzy analysis.

Fig. 6

Comparison between power utilized graph with direct and fuzzy method.

The crisp value of the input with respect to the output is shown in Fig. 7. The estimated value as per the fuzzy simulations can be seen in Fig. 7. From Fig. 6 it can be seen that a value of blood sugar (After eating) of 262 mg/DL, blood pressure (systolic) at 122 mm/Hg, Blood pressure (diastolic) at 93.9 mm/Hg and cholesterol level of 248 mg/DL the

Fig. 7

Rule viewer of the fuzzy simulations.

output medical treatment required lies in the range of active monitoring.

The de-fuzzifer process consist of calculating the crisp value from Fig. 7 by using MAMDANI model and comparing it with the simulated values. The values are calculated by initially calculating the membership function values using the input crisp values. The membership function values from the four input values are,

Blood Sugar (After eating)

u₁ = 300–262/300 = 0.13 mg/DL

u₂ = 1–0.13 = 0.87 mg/DL

Blood pressure (systolic)

u₃ = 190–122/190 = 0.36 mm/Hg

u₄ = 1–0.36 = 0.64 mm/Hg

Blood pressure (diastolic)

u₅ = 130–93.9/130 = 0.277 mm/Hg

u₆ = 1–0.277 = 0.723 mm/Hg

Cholesterol level

u₇ = 300–248/300 = 0.17 mg/DL

u₈ = 1–0.17 = 0.83 mg/DL

Combination of these membership function values are calculated. Using MAMDANI formula the output medical treatment required value is calculated using the formula below,

$MAMDANI Formula = \sum (Mi \times Si) / \sum Mi$ (1)

Where,

Mi is the membership function value.

Si is the singleton value calculated by the output state of the selected input in range of 0-1.

Table 2 shows a comparison between the simulated values and MAMDANI calculated values. The MAMDANI model calculated value is 0.47 and simulated value from Fig. 7 is 0.5. Both the simulated and calculated value lies in the active monitoring range for value of blood sugar (After eating) of 262 mg/DL, blood pressure (systolic) at 122 mm/Hg, Blood pressure (diastolic) at 93.9 mm/Hg and cholesterol level of 248 mg/DL.

Table 2

Comparison of simulated and MAMDANI model value

Simulated Value	0.5
MAMDANI model calculated Value	0.47
Error	0.03

4 Conclusion

In this work, IoT based health care data driven system has been introduced using fuzzy logic interference tool. Four inputs including the blood sugar level, blood pressure (Both systolic and diastolic) and cholesterol level are studied and on the basis of these four, rules are selected. The rules are selected on the basis of real time data and human knowledge. The data send to the healt-care provider is based on the active and emergency medical treatment require. This not only results in the decrease in power utilized by 8 times as compared to the conventionally used method. The comparison between the fuzzy simulated value and MAMDANI calculated value is less than 1% which shows the accurate of the proposed system.

References

Whitmore

, Agarwal

and Da Xu

, The Internet of Things—A survey of topics and trends, Inf Syst Front 17 (2015), 261–274.

Al-Turjman

, Zahmatkesh

and Shahroze

, An overview of security and privacy in smart cities’ IoT communications, Trans Emerg Telecommun Technol, (2019), 1–19.

Farooq

M.S.

, Riaz

, Abid

, Umer

and Bin Zikria

, Role of iot technology in agriculture: A systematic literature review, Electron 9 (2020).

Adamides

, et al., Smart farming techniques for climate change adaptation in Cyprus, Atmosphere (Basel) 11 (2020), 1–17.

Barriga

J.J.

, et al., Smart parking: A literature review from the technological perspective, Appl Sci 9 (2019).

El Zouka

H.A.

and Hosni

M.M.

, Secure IoT communications for smart healthcare monitoring system, Internet of Things (2019), 100036.

Sokullu

and Çetin

H.E.

, Internet of Things Healthcare and patient monitoring using IoT, Internet of Things 11 (2020), 100173.

Hameed

, Bajwa

I.S.

, Ramzan

, Anwar

and Khan

, An Intelligent IoT Based Healthcare System Using Fuzzy Neural Networks, Sci Program 2020 (2020).

Malarvizhi

, Lokesh

, Varatharajan

and Chandra

, Cloud and IoT based disease prediction and diagnosis system for healthcare using Fuzzy neural classifier, Futur Gener Comput Syst 86 (2018), 527–534.

10.

Selvan

N.S.

, Vairavasundaram

and Ravi

, Fuzzy ontology-based personalized recommendation for internet of medical things with linked open data, J Intell Fuzzy Syst 36 (2019), 4065–4075.

11.

Mahesh

and Harigovindan

V.P.

, Fuzzy based optimal and traffic-aware restricted access window mechanism for dense IoT networks, J Intell Fuzzy Syst 37 (2019), 7851–7864.

12.

Agana

M.A.

, Ofem

O.A.

and Ele

B.I.

, Springer International Publishing, A framework for a fuzzy smart home IoT e-health support system 69 (2020).

13.

Mumtaj

S.Y.

and Umamakeswari

, Neuro fuzzy based healthcare system using IoT, 2017 Int Conf Energy,Commun Data Anal Soft Comput ICECDS 2017, (2018), 2299–2303.

14.

Neeralagi

R.R.

, IoT based Health Monitoring using Fuzzy logic, Int J Comput Intell Res 13 (2017), 2419–2429.

15.

Tayyaba

, Ashraf

M.W.

, Alquthami

, Ahmad

and Manzoor

, Fuzzy-based approach using iot devices for smart home to assist blind people for navigation, Sensors (Switzerland) 20 (2020), 1–13.

16.

Islam

, Rahaman

and Islam

, Development of Smart Healthcare Monitoring System in IoT Environment, SN Comput Sci 1 (2020), 1–11.

17.

Satpathy

, Mohan

, Das

and Debbarma

, A new healthcare diagnosis system using an IoT - based fuzzy classifier with FPGA, J Supercomput 76 (2020), 5849–5861.

18.

Vossen

P.H.

and Ajit

, Fuzzy Scoring Theory Applied to Team-Peer Assessment: Additive vs. Multiplicative Scoring Models on the Signed or Unsigned Unit Interval, Adv Intell Syst Comput 1222 (2020), 84–111.

19.

Miramontes

, Melin

and Prado-Arechiga

, Fuzzy System for Classification of Nocturnal Blood Pressure Profile and Its Optimization with the Crow Search Algorithm, Soft Computing Applications, Proceedings of the 8th Intenational Workshop Soft Computing Applications(SOFA-2018) (2021), 23–34.

20.

Ramya

, Teekaraman

and Ramesh Kumar

K.A.

, Fuzzy-based energy management system with decision tree algorithm for power security system, Int J Comput Intell Syst 12 (2019), 1173–1178.

21.

Chakravarthi

M.K.

and Venkatesan

, Adaptive type-2 fuzzy controller for nonlinear delay dominant MIMO systems: An experimental paradigm in LabVIEW, Int J Adv Intell Paradig 10 (2018), 354–373.

22.

Iqbal

, Zhang

, Arif

, Wang

and Dicu

A.M.

, A Comparative Study of Fuzzy Logic Regression and ARIMA Models for Prediction of Gram Production, Soft Computing Applications, Proceedings of the 8th Intenational Workshop Soft Computing Applications(SOFA-2018) (2021), 288–298.