AI Connects the Dots: Using Data to Better Manage Chronic Illnesses

This article was originally published in Thinkers & Makers, a magazine from Akkodis featuring the smartest minds and innovative projects that are driving the future of technology and engineering.

The STAYWELL project (Sustainable Technology to Analyze Wellness Among Chronically Ill Patients), led by AI and IoT expert Asma Gasmi, is developing a wearable device for people with chronic conditions. The device is equipped with sensors that collect real-time data about the user's physical condition. The sensors cross-reference the data with the user's medical history and present the data to doctors and other caregivers.

One early area of focus is diabetes.

Detecting Drifts 

STAYWELL aims to detect drifts in health metrics that could point to worsening a patient’s condition.

That could be sleep patterns or heart rate. Still, in the project's early phases, the team focuses on measurements, including levels of sugar and glutamine in the blood of patients with diabetes. Glutamine helps to control sugar levels. Gasmi, who is managing the STAYWELL project and acting as lead tech in AI and IoT, explained this approach.

She has a PhD in IoT and AI, in addition to a degree in electromechanics and systems control engineering. Her background in medical technology research helps her understand what is at stake and how new technologies can help.

“To begin with, we are focusing on diabetic patients. Among other things, we measure sugar and glutamine in their blood. In particular, we are concentrating on monitoring type 2 diabetes patients. That diabetes type is often not treated properly, because many patients discover it too late.”

The principle of an early warning system could potentially help patients avoid complications, reduce unnecessary procedures, and reduce the time they spend in hospital.

And AI has an important role to play.

“AI is particularly adept at connecting the dots between health patterns and chronic illnesses,” Gasmi said. “For individuals with chronic illnesses, even minor deviations from the norm, such as changes in sleep patterns, could indicate underlying issues.”

The AI program analyzes patterns in the data collected by the wearable device to create a model of the patient. In addition to detecting and alerting deviations in a patient’s real-time condition compared to the baseline, the program uses a predictive algorithm. If the same anomalies persist, this algorithm can indicate the patient’s future condition.

Gasmi explained that AI can help the project’s team members work smarter. “As an engineer, it is not only about how much you know but also how effectively you can integrate AI into your work. I believe AI will become a part of our daily routines, saving us a significant amount of time. For example, I use AI to help me find errors in my code and to prepare elements for my presentations, allowing me to focus more on truly challenging tasks.”

Asma Gasmi, STAYWELL Project Manager, AI & IoT Lead, Akkodis (on the left)

Data Potential 

Even though the team could see the potential for data and AI to build a smart solution straightaway, achieving it was challenging. Data is a key element in the STAYWELL solution, but health data is sensitive, well-protected, and initially difficult to access.

Through Gasmi’s work with key institutions and the strength of her concept, she secured permission to work with real data and begin modeling it into a patient for study.

That hard work is paying off.

“It was challenging to establish the causality between real-time monitoring data and the risk of a patient falling into a diabetic coma, but we managed it,” Gasmi said. “Currently, our precision is at 74 %. That’s promising, and I’m sure it will improve further.

“I believe AI will become a part of our daily routines, saving us a significant amount of time.” Asma Gasmi, STAYWELL Project Manager, AI & IoT Lead, Akkodis

Non-invasive Sensors

STAYWELL builds on Gasmi’s PhD in medical technology, which was funded by a company that produces hospital beds. This company sought to develop a smart bed to analyze a patient’s sleep efficiently and non-invasively. Unlike conventional diagnostic tools in sleep medicine, which require multiple sensors attached to the brain and body, the smart bed gathered similar sleep data using noninvasive sensors.

Gasmi achieved 94% precision in data collection during her studies compared to common polysomnography (sleep study) tools. On top of that data, she developed an AI-powered prediction algorithm for elderly people living in nursing homes.

An Extra Tool for Doctors

“Building on my thesis, I decided to develop technology to make life easier for people with chronic illnesses and their caregivers. So, I came up with the STAYWELL project,” Gasmi said.

The STAYWELL project is also investigating other applications. The team is partnering with a hospital in Calais, in the north of France, to find a way to use monitoring data from the STAYWELL device for other pathologies. These include post-cancer treatment care.

 “A sensor system connecting real-time monitoring with patient history and predictive capabilities makes sense in numerous areas. Giving doctors and caregivers such a tool will improve their work and help patients,” Gasmi said.

The first working prototype of STAYWELL’s device will be ready at the end of 2024, while the ultimate proof of concept is planned for the end of 2026. After that, plans to commercialize STAYWELL will get underway.