Getting your Trinity Audio player ready...
|
Artificial intelligence (AI) is poised to change the screening and early identification of dementia, a devastating disease that affects millions worldwide. AI is being used in a big project in Singapore called Project Pensieve to look at drawings and spot early signs of dementia in users.
The Government Technology Agency (GovTech) has created a digital screening tool in partnership with the Department of Psychiatry at Singapore General Hospital that uses machine learning (ML) technology and clinical knowledge to assess dementia risk in under 10 minutes. This innovative programme is a part of GovTech’s ongoing efforts to use AI to improve healthcare, identify dementia early, and help seniors enjoy more productive lives.
As Singapore struggles with an ageing population, dementia is becoming an increasing problem. Dementia now affects one in ten older adults in Singapore, and by 2030, this number is expected to drastically increase and affect up to 152,000 people. Tragically, up to 70% of dementia cases in Singapore are not recognised until they have progressed, underscoring the urgent need for improved and scalable screening methods.
Traditional cognitive assessments, such as the Mini-Mental State Examination, are helpful in the diagnosis of dementia but are frequently labour- and time-intensive and reliant on the subject’s literacy abilities. However, drawing examinations don’t require literacy, which makes them more inclusive and accessible.
The digital tablet-based drawing tests from Project Pensieve can be completed on their own in five to ten minutes, and AI will score the results right away. By allowing non-experts from the community to coordinate screenings and reach a larger demographic of elders, this strategy fills the resource gap in dementia detection.
There were many important technological issues in the creation of Project Pensieve. The project’s AI engineers processed participant stroke sequences using a variety of methods, including meta-learning, long short-term memory (LSTM), and gated recurrent units (GRUs).
The AI can identify unusual patterns or prolonged hesitation, which may be signs of cognitive impairments, by examining these sequences. With precision anticipated to increase as more data is gathered for the continuing study, initial results showed inference accuracy rates of over 80%.
The creation of Project Pensieve placed a high priority on user-friendliness, especially for older users with little prior computer expertise. To make sure the app was understandable and accurate, user experience research was carried out, considering comments from doctors and patients. To maximise usability and readability for seniors, the design team carefully chose the tablet size, stylus type, and user interface features.
The development process for Project Pensieve includes citizen co-creation and involvement. It also takes a human-centred approach. The target demographic’s demands and preferences are met by the tool through public involvement through platforms and events.
While Project Pensieve is a big step forward in dementia screening, it’s important to keep in mind that AI should support, not replace, a qualified diagnosis. People who have been detected through AI testing as possibly having dementia should be referred to skilled doctors for in-depth evaluation and intervention.
As more data becomes available, the sensitivity and specificity of the AI will continue to be adjusted, providing accurate findings, and reducing false positives and negatives. This strategy emphasises the careful and appropriate integration of AI into healthcare.
One example of how AI is changing healthcare is Project Pensieve. In addition to dementia screening, AI is being used to decipher imaging results, guarantee medication compliance, and spot unusual conduct suggestive of underlying medical conditions.
The Project Pensieve team is actively investigating the more extensive uses of AI in healthcare, highlighting the enormous potential of this technology to improve the well-being of people and communities.