Zoe Care: using artificial intelligence to watch over the elderly

Zoe Care emerged from the meeting of science and entrepreneurship, when lecturer Piotr Antonik and entrepreneur Thomas Saphir were introduced through CentraleSupélec's technology transfer unit. At the time, Piotr Antonik was working at the Optical Materials, Photonics and Systems Laboratory (LMOPS - CentraleSupélec/Univ. de Lorraine) on various movement recognition technologies using neural network algorithms. "He noticed that this worked particularly well with Wi-Fi," recalls Thomas Saphir, himself a CentraleSupélec alumnus. At that stage, the technology was not intended for the healthcare sector or ensuring independent living for the elderly.
However, for Thomas Saphir, who brought twenty-five years' experience as a high-tech entrepreneur, the potential of this technology was clear immediately: "I was looking for a project with real impact. From my own personal experience of being a carer for my grandmother who fell quite frequently, this idea struck a personal chord. More generally, it addresses major environmental and societal challenges," explains the entrepreneur. The venture began officially in July 2022, when the pair founded Zoe Care.

A smart, discreet fall detector

The start-up is developing a minimum viable product (MVP) called ZoeFall. This fall detector uses artificial intelligence (AI) technology to identify when someone has fallen by analysing how their movements disrupt Wi-Fi signals in the home. The system immediately sends an alert to a designated contact to come to the person's aid. 
The device, which is similar to a Wi-Fi repeater, is compact and discreet. It houses a microcontroller and an integrated Wi-Fi antenna, and plugs into a power socket within reach of the home's Wi-Fi network. From there, it continuously analyses the Wi-Fi signals it receives. "It can even be hidden behind a television, because Wi-Fi waves bounce off walls and move around obstacles, unlike cameras, which need to be carefully positioned," adds Thomas Saphir. Since the signals can also pass through certain objects, including the human body, which contains a lot of water, they are distorted. These distortions vary over time depending on the body's movements. The challenge lies in the ability of ZoeFall's AI to interpret movements based on these variations.
The AI developed by Zoe Care has been trained to recognise specific movements, such as falls. However, it can detect a wide range of body movements including walking, sitting down, standing up, lying down, breathing patterns during sleep, as well as gestures related to eating or drinking. "It can detect whether someone is present in a room and give their approximate location," explains Thomas Saphir. 
Another advantage is its energy efficiency. "Today, we hear a lot about generative AI and deep learning, with models relying on powerful cloud-based data centres. In contrast, our AI is economical and embedded on a simple microcontroller, similar to those found in home automation devices such as remote controls or switches," explains the start-up's co-founder.

Encouraging early trials

The first pilot installations of the tool are now underway in residential care homes for dependant elderly people (Établissement d'hébergement pour personnes âgées dépendantes - EHPAD). "From a commercial standpoint, it's relatively straightforward to deploy, since every room can be fitted with a ZoeFall unit. It's also simple to manage the alerts, as trained care staff are present around the clock," says the entrepreneur. Zoe Care plans to bring the device into private homes by the end of 2026.
Primarily designed to support older people, the tool is also intended to assist those who care for them. "We've been working in the silver economy and medical and social ecosystem for three years now. We're in regular contact with geriatricians, nurses, care home managers and representatives of the relevant authorities. These people have been looking for a reliable solution for falls for a long time," notes Thomas Saphir.
According to the French Ministry of Labour, Health, Solidarity and Families, by 2030, France will have 20 million people aged over 60, which is a third of the population. Many of them will want to remain independent and continue living at home. This is already the case for over 92% of people aged over 70. Satisfying these ambitions is both a major challenge and a societal responsibility.
The market has seen a wave of new tools aimed at addressing this need in recent years. Currently, the most common solution is bracelets or watches fitted with sensors such as accelerometers. "One million units have been deployed in Europe, but independent testing shows a sensitivity rate of around 30%, which is very low," says Zoe Care's co-founder. In contrast, ZoeFall has achieved 99% accuracy in laboratory conditions and early field results are highly promising. 
While there are other, more recent solutions using video, audio or millimetre waves, these technologies are often expensive, sometimes intrusive and require complex installation. Consequently, they are still mainly confined to medical or hospital environments. "There still isn't a satisfactory solution available for the general public or care homes," concludes Thomas Saphir.

Anticipating needs to provide better care

As well as issuing alerts in the event of a fall, Zoe Care is now aiming to help prevent loss of independence so that older people can remain safely at home for as long as possible. "We're working on new applications focused on prevention and monitoring. This means detecting the earliest signs of dependency, so that corrective action can be taken (physiotherapy, nutrition, medication, etc.) in partnership with healthcare professionals," says Thomas Saphir.
To turn its ideas into reality, the start-up is drawing on a strong network of partners. From the outset, the 21st by CentraleSupélec accelerator has supported Zoe Care by providing workspace, strategic guidance from its teams, help with funding applications and event organisation. "They also encourage networking with other start-ups," adds the entrepreneur.
But that's not all, the Zoe Care team has also benefited from an initial support programme by SATT Paris-Saclay. It is set to renew this partnership through the Tech Transfer Program to develop a medical device based on its technology and identified uses.
Finally, as part of its maturation phase, Zoe Care has also forged a partnership with the University Research Laboratory in Automated Production (Lurpa - Univ. Paris-Saclay/ENS Paris-Saclay).
With a €500,000 fundraising round currently underway, Zoe Care's two co-founders are planning to launch the Zoe Fall MVP on the B2B market this autumn, targeting care homes, while also embarking on the next phase of maturation focused on R&D.