Silo AI and Verso Vision develop computer vision to assist healthcare staff

Verso Vision, a leading AI-based solution provider for fall prevention and detection for healthcare, and Silo AI, one of the largest private AI labs in Europe, have teamed up to accelerate AI development.

Verso Vision uses proven AI to prevent falls before they happen and help healthcare staff respond to them faster when they do. Together with Silo AI, Verso Vision has been able to create a light-weight, yet robust, computer vision solution that enables healthcare providers to improve care personnel’s response time and reduce patient falls in hospitals and wards. During the past three years, Silo AI has contributed through experience and expertise for specific AI challenges, and more generally helped Verso Vision build their own in-house AI capability.

Real-time and scalable, while conserving privacy

The core problem Verso Vision set out to solve is simple: prevent falls in care facilities, and help healthcare staff respond to them faster when they do. Verso Vision’s novel computer vision solution does just that. The solution signals healthcare staff whenever it believes that a patient has fallen and needs assistance. Patient rooms are automatically monitored 24/7 by machine learning models, which in turn output signals to dashboards and alarm systems. No data is stored at any stage of the loop, meaning that patient privacy remains at the center of the process.

Scalability is of the essence in Verso Vision’s solution, as it in some cases is required to monitor hundreds of rooms and patients. Silo AI’s team supported Verso Vision in developing the deep learning models used for detection, e.g. multi-person detection, pose estimation and action recognition. The main challenge was balancing limited local processing power and sufficient real-time accuracy. The solution is often set up locally, although cloud computing is also an option.

Katja Taalas, CEO and Founder of Verso Vision says:
“For us, it has been extremely useful to work in partnership with Silo AI to design, build and implement the Verso Vision AI solution. Silo AI’s experience in specific machine learning tasks has complemented our own AI expertise and helped us build a privacy-first computer vision solution that enables safer and more reliable health care.”

The computer vision pipeline features a multitude of components that are constantly developed and improved based on testing and feedback from hospitals and nursing homes. All machine learning models are tailored to the task since off-the-shelf AI components rarely work in real and complex care environments.

"Verso Vision has been using machine learning in their product early on and their AI models have gone through several cycles of improvements through the years of development and testing. With Silo AI's help, Verso Vision has been able to speed up AI development and leverage cutting-edge advances in deep learning in their product. It has genuinely been a pleasure working with such a skilled and great team.”, comments Niko Vuokko, CTO at Silo AI.

Universal solution with easy implementation

Verso Vision’s solution is built to be generic, and thus to suit any healthcare provider anywhere. The implementation time of the solution is not affected by the amount of rooms or beds. The models learn and adapt to any specific room and situation details, while keeping patient privacy at the core.

The solution offers significant value to both patients and healthcare providers. By continuously analyzing the patient’s movement and surroundings, the solution’s AI algorithms are able to reliably predict falls before they happen, prevent patients with dementia wandering out and getting lost, greatly influencing the patients’ quality of life and care. The solution lets nursing staff focus better on their job and respond to patients faster when needed, reducing nurses workload. This helps with patient safety, especially in understaffed hospitals and care homes.

“Verso Vision is building cutting-edge technology, relying on both traditional and modern computer vision, in addition to supporting a significant cause in healthcare. The commitment to their honorable purpose – caring for those who have cared for us – can be seen in their entire team, and it is always great to join forces in building AI for causes like this.”, says Peter Sarlin, CEO at Silo AI.

Results

• Verso Vision and Silo AI developed deep learning models suited for specific case and hardware requirements
• Together Verso Vision and Silo AI built a solution that saves countless work hours and makes sure that patients receive care before and on time
• A universal solution that runs locally, is light-weight and accurate, and follows strict privacy requirements