Singapore has one of the best healthcare systems in the world, but it is still facing challenges.

Its population is aging and older adults are experiencing more chronic diseases. Lifespans are growing, but healthspan – the years in your life where you are healthy – is not keeping pace. Like any healthcare system, Singapore doesn’t have unlimited financial or personnel resources. It has to get creative. 

One of the agencies looking into tackling these healthcare challenges is MOH Office for Healthcare Transformation, a group set up by Singapore's Ministry of Health in 2018 to address Singapore's healthcare challenges by leveraging innovation and technology.

Its work in artificial intelligence stands out as an important case study for any healthcare system looking to use technology to address a shortfall.  

Over the past six years, MOHT and Singapore’s Institute of Mental Health have together piloted various remote patient monitoring programs powered by AI that examine a span of chronic conditions and mental health. Patients use sensors such as blood pressure monitors, fitness trackers, and smartphones to track their health as they go about their lives. Patients then receive monitoring tips, sometimes coming from AI, and sometimes from a care worker.

In one of the projects focusing on serious mental health conditions such as schizophrenia, called Project HOPES, machine learning is used to analyze digital signals collected via wearable devices. A mobile app then helps users to self-manage their condition, as well as keeps them connected to a health provider. HOPES even detects behavioral patterns – like heart rate and sleep behavior – before symptoms become apparent. 

The group has launched mindline.sg, where Singapore residents can directly access hundreds of mental health resources, interact with an emotionally-intelligent AI chatbot, and, when necessary, connect directly with mental health professionals. About 20% of Singapore’s population has visited the mindline.sg platform.

It’s worth pausing to reflect on the lessons learned. 

First, Singapore realized early on that generative AI may only be used as a support tool for clinical care decisions, and all medical and patient care decisions continue to require human oversight. A more cautious approach is needed, beginning with explainable, rules-based methods, where algorithms designed by people perform specific, well-defined tasks.

Generative AI is used to listen and recommend, such as giving hints to a therapist who is counseling a patient on the phone or via text messaging. And the therapist is trained to take responsibility for everything that is relayed back to the patient. Generative AI does not yet write back or interface the patient directly.

It’s a more cautious approach and likely a safer one when we are learning in real time about the limits of AI and mental health. 

One of the biggest takeaways from Singapore is that you don’t need to use a wearable to be monitored — your phone alone can provide a wealth of information. Smartphones can collect dozens of different digital signals that can be analyzed to identify certain behavioral patterns. For example, if you want to monitor sleep, how often someone picks up their phone during the night can serve as a useful indicator.

“The phone can be used quite effectively to approximate sleep duration, since the last thing most people do before bed is use their phone, and the first thing they do after waking up is pick it up again,” says Robert Morris, Chief Technology Strategist for the health innovation office. (For further reading, Robert has recently finished a book outlining the key ways AI is likely to transform healthcare in the coming years.)

Other indicators being monitored, and which researchers hope will yield insights over time, including how often someone leaves their home and even how quickly they type on their phone. For example: Does slower typing than usual signal stress? Or, does a decreased number of steps indicate poor mental health?

“When it comes to HOPES, we think we might have the largest dataset of digital biomarker events collected on psychosis in the world — what researchers call a ‘digital phenotyping’ dataset,” says Thisum Kankanamge, chief engineer of the HOPES project at MOHT.

Thisum adds that HOPES doesn’t just help doctors intervene early, it also nudges patients by empowering them with timely information and tools.

“For example, if a patient’s sleep pattern has been significantly disrupted over the past few days, the algorithms will detect it and alert the doctor, while also offering the patient self-help tools such as sleep exercises, relevant cognitive behavioral therapy resources, and educational articles,” added Thisum.

To be sure, this type of tracking is not possible in America. Americans don’t trust technology and we absolutely don’t trust the government enough to let them inside our worlds this much. 

But this tracking will provide insight for American doctors, who are watching the developments in Singapore closely. And it could be a blueprint for how doctors can use large language models in all types of health care. Not as a replacement, but as a tool to reduce the time to find optimal therapies or provide more real-time feedback in care. 

Not to mention, you can take some of this into your own hands. It’s easy to buy a fitness tracker or smart watch and uncover your own health insights. The specialists in Singapore say residents there are already doing so.

“Patients don’t want all of the data,” says Thisum. “They might just want information on sleep.” And that’s the point: when technology is used thoughtfully, even small insights can empower smarter choices.

Over the course of the past year, the team at Bloomberg has published a series of revelatory, concerning, and eye-opening reporting about the business of cancer.

There have been stories about why cheaper drugs might be better for patients, a can’t miss graphic about the cost of treatment, and this week, a story about the gold rush for treatments Keytruda and Avastin.

The most relevant story for survivors might be this one about a BioZorb device that supposedly didn’t dissolve. My reaction to this story is simple: Even if you are years out of treatment and well past any worries about cancer, still see your oncologist.

Every week, doctors are finding out new information about old treatments — be it devices, chemotherapy, or otherwise — and communicating with your oncologist is the best way to find out if these findings are relevant for you.