From data to healing: The potential of analytics and AI in advancing mental health

Kitty Lee
Managing Director, Healthcare Transformation,
Temus

One of our greatest global health challenges is mental health. As this challenge increasingly becomes more dire in recent years accelerated by COVID, many digital solutions have arisen. Many of these have put data science and AI, and most recently generative AI and Large Language Models, at their core. I presented three ‘calls’ that are thematics of what will drive real-world application in mental health and some examples of global innovators that are examples of these themes. These were presented alongside speakers from MOHT and SAS Institute at a recent IDEAS event, hosted by Temus and organised by the Singapore Computer Society (SCS).

Real-world applications of AI in mental health can:

1. Enable precision in mental health. ML and AI can help more accurately diagnose and provide personalised treatment plans by collecting and analysing a wide range of data points (e.g., medical history, symptoms, and behavioural patterns) instead of relying on traditional inputs such as self-reporting.
2. Combine physical and mental wellbeing together. Mental health has physical indicators that offer a far more powerful, objective basis, especially when we combine AI with other technologies such as IOT.
3. Build solutions that provide immediate cost savings now, instead only solutions that provide potential savings in the future over very large population cohorts.

Bright spots in the industry:

One example AI being employed in digital health that I spoke about was Spring Health from the US. Their precision mental healthcare approach involved clinical assessment, personalised care plan, real-time provider feedback & recommendation and digital content. Their mobile and web platforms use machine-learning models leveraging millions of data points to tailor interventions and treatment plans. Users reported an improvement in mental health (~70% of participants in a study improved their mental health), with fewer missed workdays, increase in productivity, and up to an average of US$7k in cost savings per participant.

In the UK, BioBeats demonstrates how users can combine physical and mental wellbeing. It is a mental health app leveraging AI to interpret sensor data such as heart rate variability and activity, as well as psychometric data from a wearable. A wellbeing score offers employees an overall measure of mental wellbeing based on personal health data e.g., sleep, activity, heart rate, mood and cognitive function; and delivers digital therapeutics. Companies saw a 31% cut in employee absence and 54% decrease in cost from reduction in length and number of sick leave when their staff used the app.

Big Health’s flagship product is Sleepio, a cognitive behavioral therapy app that aims to help users’ poor-quality sleep and insomnia. They use an artificial intelligence (AI) algorithm to provide people with tailored digital cognitive behavioural therapy for insomnia (CBT-I). Evidence is everything in digital transformation: Big Health designed an interrupted time series analysis, comparing primary care use before and after the rollout of Sleepio, and focused on how many times people saw their GP and the relevant prescriptions they received. Sleepio became the first digital therapeutic to receive NICE (National Institute for Health and Care Excellence) guidance for NHS use last year to treat insomnia.

So, why are we not already using technology widely to enhance healthcare?

Technologists must collaborate with both medical practitioners and financiers to ensure the implementation and effectiveness of the solution. Because medical treatments will inevitably involve insurance and/or public subsidies, there is also the question of whether digital care or prevention can pass the traditional lens and evaluation criterion to enable coverage. Lastly, verdict is still out on what is the ‘right’ balance between safety and innovation. We are pioneering some of these technologies and the infrastructure & processes of industrialising technology hasn’t caught up yet. This includes the necessary medical and legal frameworks for endorsing & measuring efficacy. Current medical and legal frameworks do not account for measuring the efficacy of AI-enabled healthcare, so tests need to be designed to ensure accuracy and robustness.

In conclusion, AI is an amazing tool when added to the healthcare toolbox, but not a silver bullet at its current stage of development. It is most powerful when combined with other technologies for a more comprehensive and practical solution. We also must recognise the barriers to adoption and scale within the sector if we hope to continue to push the boundaries of health.

AI is here to stay, and here to help. The brightest brains around the world are already pouring capabilities and resources into this field, and we are only just starting to see some of these solutions bloom.