An alert and ranking system to identify high-risk patients

In healthcare, risks to patients are common and it is very crucial to be one step ahead of incident, where the life of a patient could depend on a quick response time and a finely tuned understanding of where something is going off. Organizations need to specifically recognize people at elevated ris

Project Title

An alert and ranking system to identify high-risk patients

Project Area of Specialization

Artificial Intelligence

Project Summary

In healthcare, risks to patients are common and it is very crucial to be one step ahead of incident, where the life of a patient could depend on a quick response time and a finely tuned understanding of where something is going off. Organizations need to specifically recognize people at elevated risks as early as possible to have the greatest chance of helping patients prevent long-term health complications that are expensive and difficult to manage. The healthcare sector is being changed by the ability to record vast volumes of patient information and to analyze the tremendous volume of data generated using diverse data processing methods that add a paradigm shift to healthcare. Thus, more and more AI applications are being developed as they enhance the opportunity for healthcare providers to truly understand needs of the patients they work for, and with that awareness they can offer better insight, direction, and support for remaining well which in turn can promote better clinical decisions. It can also be inferred that using AI can eliminate failures that are inherent in human clinical practice. Our project goal is to design a wearable device along with software interfaces that can help save high-risk patients' lives and identify them. It will use advanced machine learning algorithms to train a model based on different Kaggle data sets to evaluate and prioritize the severity of heart patients. We will primarily focus on this area and might later work on other domains as well. Heart disease is the most leading cause of death and therefore we have chosen this domain to work with initially. Cardiologists, these days use numerous imaging tests and invasive blood pressure measurements to investigate and track the seriousness of these diseases, so they can keep this project in their toolbox to assist them in their daily work. It would also be able to measure patients' temperature, heart rate and blood pressure using devices with sensors which will also be developed as the part of the project. The model will also be used anytime anything suspicious happens to alert the administrators. For instance, if a patient's heart rate increases, it denotes something serious and administrators will receive SMS or notification on their app to take some immediate action. Along with this, it would have ample built-in knowledge to prescribe some medications based on the questions answered by the patients such as chest pain, fatigue etc.

Project Objectives

• Promoting the paradigm of proactive health care.
• Promoting early diagnosis which would lead to rapid action and ultimately faster recovery.
• Determine the severity of illness in patients based on symptoms.
• Recommend drugs/medicines for common problems.
• Check temperature, blood pressure via sensor devices.
• Notify doctors/administrators via SMS or any other notification system in case of an emergency.
• Develop a wearable that will monitor the patient and send data to our server and patient’s current state will be analyzed.
• Develop a web and a mobile application with a user-friendly interface to be used by administrators/doctors.
• Increase patient safety.

Project Implementation Method

Our project is an integrated framework for HW/SW. It is focused solely on advanced machine learning algorithms to classify patients with high risks. To obtain valuable data from a broad patient population to help make real-time inferences for health risk forecasts with precision, our model will be trained using different datasets from the Kaggle.

Our wearable devices will be custom-built (powered by ATmega8 microcontroller) and collect data from the subject every three seconds via sensors and send it for further analysis to the server. Every wearable will have a unique Id to be able to distinguish from other devices, and this unique id will be assigned against one patient, this single device can later be used by another patient, and the past data of the patient before will be moved to cold storage. These will communicate through the Internet to the server, which will be operated by a Wi-Fi module. The information is transmitted using the HTTPS protocol to ensure the security of the patient's confidential data at the network level. Each http request will be accepted by the server and stored in the database and forwarded to the machine learning algorithm for severity analysis. Each http request will be accepted by the server and stored in the database and forwarded to the machine learning algorithm for severity analysis, earlier disease detection, risk prediction and customized treatment plans to ensure optimum performance. It will influence the performance of hospitals and health services, thus reducing the cost of treatment and informing patients about alternative disease routes.

The machine learning service responds back with severity in percentage 100% being the highest 0% being the lowest, the exact mild percentage will be decided after training the model on test data. If machine learning service responds back with a severe percentage, the API server will raise alerts using Push Notifications for web and mobile, Email, SMS, and automated phone calls.

We would have both a web application and a smartphone application to be used and tracked by administrators of such patients, as the system is HW/SW integrated. DynamoDB will be used as the system's database. The solution will be deployed at Amazon Web Services which ensures the system has 0% downtime, as some patient's lives depend on this.

Benefits of the Project

Wearable devices are changing the healthcare environment at a remarkable rate, creating a whole new digital medicine landscape. They promote the opportunity to obtain prompt medical assistance. Our system can help healthcare providers track patient information over a long period of time and have a deeper understanding of the conditions that concern patients. For instance, our technology will be used to track them at home to guarantee that no complications arise if the patient is at risk but not critically ill enough to be in the hospital. This will be a groundbreaking progress for both healthcare specialists and patients.

For both patients and healthcare professionals, it will be very user-friendly and easy to use so that no technological assistance is required that saves money and time. Our notification system will be very efficient in a way that if anything goes off, administrators/doctors will get SMS on their devices. Administrators will then review the system immediately to determine whether the help is needed. Allocation of limited medical supplies and resources would get easier as we classify risk according to the early diagnosis. This would also give a lead and time to prepare to the people involved in healthcare as well as the end user which also promotes fast recovery.

It can also help encourage patients to improve their health by helping them monitor their progress through wearables that, in turn, promote self-improvement and keep track of their rewards.

As the system also offers advice for common conditions, doctors' productivity would be improved. Other than that, digitally monitoring blood pressure and temperature instead of manual instruments often saves time. This would decrease the burden on medical facilities which are scarce in our country.

Using machine decision support to help classify high-risk patients and alert their administrators immediately results in increased patient safety which can become a very valuable method to help healthcare providers and save lives, which is humanity's prime aim.

Technical Details of Final Deliverable

The following things will comprise our final items:

• Hand band with pulse sensor, temperature sensor and ATMega8 microcontroller and Wi-Fi module with rechargeable Lithium-ion 2000mAh battery.
• Hardware systems for temperature and blood pressure tests.
• Web application to be used by the administrators.
• API server powered by Node.js that will accept requests over HTTP hosted on Amazon Web Services EC2
• Machine learning service powered by Python, that will also accept data over HTTP hosted on Amazon Web Services EC2
• API server connected with Plivo for SMS and Phone call automation
• API sever connected with Amazon SES to send emails
• API server connected with Firebase for web and mobile push notifications
• User Interface created with React connected with API server
• API server connected with Amazon DynamoDB database
• Mobile Application created on Flutter connected with API

Final Deliverable of the Project

HW/SW integrated system

Core Industry

Health

Other Industries

Core Technology

Artificial Intelligence(AI)

Other Technologies

Internet of Things (IoT), Wearables and Implantables

Sustainable Development Goals

Good Health and Well-Being for People

Required Resources

If you need this project, please contact me on contact@adikhanofficial.com