Applying AI technology to one of the main concerns in patient safety, Nanyang Polytechnic (NYP) and local medical technologies firm Longway AI Technologies have developed a system that can analyse human gaits and make recommendations for healthcare professionals to render assistance before a fall happens.

In Singapore, data indicates that one-third of people aged 65 and above have suffered a fall at least once, with 40 per cent of these falls accounting for injury-related deaths. 30 per cent of these individuals will experience recurring falls. Older individuals are more likely to have a fall due to physical or medical reasons, and gait change is a critical early sign of a potential fall.

In the US, an article published in the Centers for Disease Control and Prevention (CDC) last year notes that mortality increases for older adults (aged above 65 years old) if they have a fall[1]. A study by the Ministry of Health here in Singapore showed that patients with hip fractures also have adverse consequences: One year post fracture, the mortality is between 20 to 27 per cent. Of the survivors, 20 per cent become semi or fully dependent, and 39 per cent say they had reduced mobility[2].

Explains Dr Phua Chee Teck, Director of the NYP School of Engineering: “Essentially, we are teaching the system to recognise unsteady gaits. The AI then takes over the monitoring and it must make a decision autonomously whether the unsteadiness has reached a trigger point so that someone must be alerted.”

During the development stage, the AI system was trained with over 200 hours of videos across two phases. The pre-training phase saw the system analyse over 200,000 open-source videos that covered training, testing, and validation. In the second phase, the AI model was further finetuned with more than 300 videos for each type of human activity – including basic activities like standing up, sitting down, and taking short walks – to allow the system to understand biomechanical movements associated with unsteady gaits.

This was done through tapping on 3D convolutional neural network (3DCNN) – an AI method that teaches the computer to process data in a way modelled by the human brain and nervous system. Through deep learning, a machine learning process that uses interconnected nodes or neurons in a layered structure that resembles the human brain, the AI system was programmed to react based on the movement of different joints from a human skeleton post model. And when the system detects specific movements with higher probabilities of leading to a fall, it sends the relevant signals to the healthcare team nearby.

Adds Dr Phua: “The secret recipe involves getting the training correct – making sure that you’re not causing unnecessary alarm with false cases, and to really be able to trigger the care professionals should the event be predicted with high certainty.”

Since the system was developed at the School’s Centre for Innovation for Electronics and Internet of Things (COI-EIoT) in 2020, various pilot trials have been conducted with partners, including Singapore General Hospital (SGH) and St. Luke’s ElderCare Rivervale Centre:

i.                     The AI system was able to help healthcare staff prioritise specific patients for immediate acute care at the A&E.

ii.                   The system enabled St. Luke’s to monitor some 40 seniors within the facility, helping to optimise manpower deployment.

Currently, an enhanced system has been deployed and trialled at the SASCO Senior Citizens’ Home at West Coast. NYP is also in discussions with St. Andrew’s Nursing Home (Queenstown), to deploy the system within their premises and explore how it can aid in other areas of their residential care.   Derrick Ng, Senior Operations Manager, St. Andrew’s Nursing Home (Queenstown), said, “We are excited to explore the system through using real-time data and predictive insights so that we can further enhance our residential care in different areas.”

Damien Ooi, Head, Centre Management, SASCO Senior Citizens’ Home, said, “We believe NYP’s AI solution will be instrumental in enhancing our care for the elderly. Their innovative approach will not only aid in immediate assistance but also contribute significantly to our overall care strategy. Through this pilot, we hope to better understand and anticipate the needs of our seniors and provide prompt assistance to prevent falls within

NYP is also exploring collaborations with the geriatrics departments of several hospitals, aiming to broaden the impact of this technology in the healthcare sector.

[1] https://www.cdc.gov/mmwr/volumes/72/wr/mm7235a1.htm  

[2] https://www.moh.gov.sg/docs/librariesprovider4/guidelines/cpg_osteoporosis.pdf