Human Arthritis Analysis and Prediction using Non Invasive Sensors
Under the JPMA study, there was only one cross-sectional study, by Khalil et al., Which reported the human and social, and economic data of RA patients in Pakistan. The study was based solely on patients living in urban areas. It was reported that 55% of patients were illiterate. In ad
2025-06-28 16:27:43 - Adil Khan
Human Arthritis Analysis and Prediction using Non Invasive Sensors
Project Area of Specialization Wearables and ImplantableProject SummaryUnder the JPMA study, there was only one cross-sectional study, by Khalil et al., Which reported the human and social, and economic data of RA patients in Pakistan. The study was based solely on patients living in urban areas. It was reported that 55% of patients were illiterate. In addition, the majority (66.5%) of patients had monthly family incomes between PKR 5,000 and 20,000, 13.5% had incomes below PKR 5,000 and only 11.5% had higher incomes than PKR 50,000. In terms of growth, 3 studies were conducted in Karachi from 2011 to 2015 that reported an increase of 12.9% in RA-affected patient count. This was followed by another study using the same method that reported the highest increase of 21.7% in the tally. In a study expected by Shamim
and colleagues reported an increase of 26.9% in patients.
All the methods and techniques discussed above are either too expensive or time-taking. To overcome these and many more shortcomings, this project aims to identify arthritis
at an early stage with bearable cost and usable by the majority of the potential patients.
To achieve this claim, we integrated several sensors to make a prototype. The purpose of the prototype is to collect data from the human knees. The collected data was stored in an SQL database and a machine learning model was created to get trained on the collected data. The machine learning model analyzed the data of patients and gave results that showed if the person has osteoarthritis or not. A website was created to run the machine learning model and show the results.
Project ObjectivesThe objectives of this project are to use non-invasive sensors for osteoarthritis diagnosis.
- To study different types of sensors and how they can be used in disease identification.
- To analyze the range of movement and vibration, and cracks and to monitor the health conditions related to the disease. integrate the motion, sound, and wearable sensors data.
- To record the data of the patients with the help of sensors.
- To apply various Machine Learning techniques to the collected data.
- To use the model for the prediction of Arthritis.
- To create an application to be used as an interface for providing patient data.
- To reduce the cost of laboratory tests.
The figure below shows the general workflow of a system that is integrating sensors to take readings and then lists down the participants on the basis of their previous medical records. which follows by collecting data from participants with the help of the Integrated sensor now. After the collection of data, different machine learning models needs to be applied to make the predictions and identification on data of the participants collected. In the end, mobile or web applications provide an interface for entering the patient data and predicting whether he/she is having arthritis.
The data is collected from the patients with the help of a hardware prototype and general information about the patient is also collected. Also, the Knee Injury and Osteoarthritis Outcome Score (KOOS) data are collected. Hence, there is data from three sources basically which are:
- General Questions about patient
- KOOS questions
- Sensor data
The proposed system is divided into three main sections as shown in the figure shown below in order to provide the optimal solution to the problem. The first step is data acquisition from the integrated sensors. In this dataset, we have applied different machine learning algorithms including Adaboost. In the end, the results and conclusions obtained from different models have been recorded and displayed with the help of a web interface.
We have performed different experiments by using different Machine learning Models. Previous work was done by using a Support Vector Machine(SVM), KNN, and Random forest. We have focused on Adaboost and SVM because of the excellent performance of these architectures in this domain.
Boosting algorithms work by converting a number of weak learners to strong learners. The goal of boosting algorithms is to build a model on the training data, and then the second model is designed to correct existing flaws in the first model. This process is continued until errors are minimized, and the data is accurately predicted. We see accuracy varies when we create a different model on the same data. But what if we use a combination of all these algorithms to make a final prediction? We will get the most accurate results by taking the results of these models. We can increase the ability to predict in this way.
AdaBoost is also known as Adaptive Boosting is a machine learning method used as an Ensemble Method which is used commonly for regression and classification problems. The most common algorithm used with AdaBoost is single-level decision trees. These trees are also called Decision Stumps. What this algorithm does is that it creates a model and provides equal weights for all data points. It then gives us high weights on points that are poorly classified. Now all the points with the highest weight are given extra value in the next model. It will keep training models until an error is minimized. The flowchart of our machine learning model using AdaBoost is shown below in Figure.
The proposed system covers a vast range of benefits for the users. It is an application of the 3rd goal of sustainability given by the United Nations that ensures healthy lives and promotes well-being for all of all ages. World statistics show that about 3 million people have arthritis globally. The statistics also show that RA is 0.55% prevalent in Northern Pakistan and 0.142% prevalent in urban communities of Karachi.
Studies have shown that the majority of arthritis-infected patients are middle-aged to old-aged. These cases are usually on stage 3 and above. The reason is that early age diagnosis is not considered by patients and by the time it is noticed, patients undergo a lot of suffering.
This system is capable of diagnosing arthritis even at an early stage so that the user can be diagnosed and the patients will be able to get treatment at a suitable time.
Technical Details of Final DeliverableFollowing are the technical deliverables of our project:
- Hardware knee-prototype to collect the data
- Database to store the collected data
- Machine learning model
- Website to show the results
| Item Name | Type | No. of Units | Per Unit Cost (in Rs) | Total (in Rs) |
|---|---|---|---|---|
| Total in (Rs) | 76528 | |||
| MPU 6050 | Equipment | 3 | 300 | 900 |
| Accelerometer MAX 30102 | Equipment | 3 | 350 | 1050 |
| Male to Male 20CM Jumper Wire | Equipment | 2 | 100 | 200 |
| Small SD Card Module | Equipment | 5 | 100 | 500 |
| DS1307 RTC Module | Equipment | 2 | 120 | 240 |
| Logic Level Sensor | Equipment | 1 | 70 | 70 |
| LED 5MM | Equipment | 12 | 1 | 12 |
| Jumper Wire Male to Female 30 CM | Equipment | 2 | 90 | 180 |
| ESP8266 Node MCU | Equipment | 1 | 500 | 500 |
| Bit 40 WATT Iron | Equipment | 1 | 80 | 80 |
| Solding Wire 100KG | Equipment | 1 | 480 | 480 |
| Gyroscope MPU 6050 | Equipment | 1 | 300 | 300 |
| Battery Clip for Arduino | Equipment | 1 | 35 | 35 |
| 9V Battery Rect | Equipment | 4 | 15 | 60 |
| Wire Cutter YH109 | Equipment | 1 | 220 | 220 |
| 9V Battery | Equipment | 2 | 60 | 120 |
| 220R 1/4 Resistor | Equipment | 12 | 1 | 12 |
| Push Button Small 4 Pin | Equipment | 16 | 2 | 32 |
| 10UF 25V | Equipment | 6 | 5 | 30 |
| 10K 1/4 Resistor | Equipment | 12 | 1 | 12 |
| 10K Volume | Equipment | 3 | 15 | 45 |
| Toggle Switch 3 Pin ON OFF | Equipment | 3 | 40 | 120 |
| Veroboard Small | Equipment | 2 | 40 | 80 |
| Node MCU ESP8266 CH340 | Equipment | 1 | 450 | 450 |
| Header Single | Equipment | 1 | 15 | 15 |
| ROM | Equipment | 1 | 400 | 400 |
| 9B4 | Equipment | 1 | 400 | 400 |
| Glue Stick | Equipment | 5 | 40 | 200 |
| Glue Gun | Equipment | 1 | 725 | 725 |
| Elastic | Equipment | 1 | 2500 | 2500 |
| Outer Covering | Equipment | 1 | 440 | 440 |
| Plier | Equipment | 1 | 350 | 350 |
| Wires | Equipment | 1 | 400 | 400 |
| UR33 | Equipment | 1 | 1050 | 1050 |
| Soldering Rron | Equipment | 1 | 650 | 650 |
| Soldering Iron stand | Equipment | 1 | 1700 | 1700 |
| De-soldering Pump | Equipment | 1 | 200 | 200 |
| Soldering Paste | Equipment | 1 | 100 | 100 |
| ECG Monitoring Sensor Module | Equipment | 1 | 2000 | 2000 |
| HK-2000B | Equipment | 1 | 15000 | 15000 |
| HK-11B | Equipment | 1 | 15000 | 15000 |
| Respiratory Sensors | Equipment | 2 | 10000 | 20000 |
| Transportation Expenses to Hall Road | Miscellaneous | 15 | 150 | 2250 |
| Transportation Expenses to Hospital | Miscellaneous | 24 | 80 | 1920 |
| Documentation Printing Expenses | Miscellaneous | 3 | 500 | 1500 |
| Thesis Printing | Miscellaneous | 1 | 4000 | 4000 |