Source: Asia Pacific Region 2 – Singapore
Navigating Health Technology
Mr Tan Tee How
Board Chairman, National Healthcare Group
Professor Joe Sim
Group CEO, National Healthcare Group
Dr David Ng
Chairman, Organising Committee of the Singapore Health & Biomedical Congress 2024
Ladies and gentlemen, friends, a very good morning
It is my pleasure to join you at the Singapore Health and Biomedical Congress.
2. Throughout human history, technological breakthroughs have always been doubled-edged swords. For example, the invention of the steam engine and spinning jenny ushered in the industrial revolution. World trade and GDP went up, but at the same time, there was oppression of workers which gave rise to Marxism, Communism, and the Cold War, which is still playing out today.
3. Closer to the present times, the Internet makes infinite amounts of knowledge accessible to everybody, but with it comes cybercrime and a threat to the mental health of one whole generation.
4. Healthcare is at the verge of a historical technological breakthrough. The unprecedented availability of data can fundamentally transform healthcare.
5. In particular, I believe the convergence of three very powerful scientific and policy forces – genomics, AI (or artificial intelligence) and the focus on preventive care – these will have profound impact on healthcare when they come together.
6. Today, I will talk about how we are going to respond to the healthcare technological breakthrough, recognising that this will be a double-edged sword.
Safeguarding the Downsides
7. Let me talk first about mitigating the key risks. I believe innovation can only take off in its fullest potential only when we know we are protected against the main risks. There are two significant ones in healthcare.
8. First, the undermining of the moral and ethical mores of society. Genetics define the core make-up of a person. Advances in science have made it possible for genetic information to be easily and inexpensively mapped out, and even for the genes to be manipulated.
9. Hence, defective genes can now be edited to treat diseases. But the same technology could theoretically be used to design babies before they are born, or select employees and grant university and medical school places, based on genetic qualities. It is not impossible.
10. Insurers can use genetic information to decide on insurance coverage – how much can you be charged and what to exclude. That undermines the whole purpose of insurance, which is to protect us against unexpected bad luck, including being dealt with a bad genetic hand at birth.
11. That is why the Ministry of Health (MOH) has worked with the Life Insurance Association to put in place a ‘Moratorium on Genetic Testing and Insurance’, which disallows the use of genetic test results for insurance underwriting. At some point, we need to strengthen the moratorium and give it some permanence.
12. MOH is therefore working on new legislation to govern the use of genetic and genomic test data. It aims to address the potential undesirable outcomes, such as the discriminatory use of genetic information in areas such as insurance and employment. We will conduct broad consultations and hope to submit the Bill to Parliament in the next one to two years.
13. The second major risk is the escalation of healthcare costs. Genomics has made precision medicine possible. This means tailoring medical treatment to the unique genetic qualities of a patient to treat serious diseases like cancer. But because this is tailor-made treatment, precision medicine can be very expensive, costing hundreds of thousands of dollars per treatment.
14. In healthcare, the language of innovation is quite different from most other sectors. In other sectors, innovation means better performance or quality, at the same or even lower price. Just look at the performance and prices of semiconductor chips, TVs, smart phones, laptops and automobiles over the decades, and we understand the value and benefit of innovation.
15. However, in healthcare, innovation often means something else. It means increasing the chances of treating a serious disease, or prolonging quality life, at a higher cost. This is the double-edged dimension of precision medicine.
16. In time, precision medicine will increasingly become mainstream clinical practice. However, I don’t think any healthcare financing system in this world is designed to fully fund precision medicine. This means that left on its own, it is very likely that only the rich will be able to afford precision medicine, leading to serious inequity in healthcare.
17. We are therefore taking steps to embrace precision medicine, and prepare for the day when it becomes mainstream and do it in a sustainable way.
18. For example, we are investing in local capabilities to develop precision medicine treatments, shorten production times and lower costs in the near future.
19. We have strengthened health technology assessment to robustly evaluate the cost effectiveness of high-cost treatments, including precision medicine. Even if the medicine increases the chance of a treatment that can prolong quality life, but costs so much more, we can conclude that it is not cost-effective. Where they are proven to be clinically and cost effective, we will subsidise these therapies.
20. The MediShield Life Council is releasing their recommendations next week. on the review of MediShield Life. They are releasing the report next week, but let me break some of the news that they are going to recommend. They have recommended for MediShield Life coverage to be extended to approved precision medicine therapies.
21. MOH intends to accept this recommendation. That way, the approved precision medicine therapies or high-cost therapies can be brought into our subsidy, MediShield Life and MediSave, or S+2M, support framework, and all Singaporeans can benefit from these therapies.
22. Mitigating the risks of technological breakthrough is often a reactive necessity, but harnessing its opportunities to transform healthcare requires proactive enterprise.
23. For the rest of my speech today, I will focus on the upside of breakthrough technology in healthcare. There are three aspects:
- First, applying AI in health institutions;
- Second, developing predictive preventive care; and
- Third, building up the IT infrastructure systems to enable these capabilities.
AI-Enhanced Healthcare
24. First, how we apply AI in hospitals and clinics.
25. Healthcare has an advantage in embracing technology because it is a highly regulated sector. Contrast this with the creative industry, where AI is almost wreaking havoc. You can use publicly available AI tools to create a song in the style of say, Taylor Swift, sung in her voice. and even create her image Or you can make a dead actor come to life. and star in a movie. All this is possible. It undermines artistic creation, and the genie is already out of the bottle. I don’t think you can bottle it back.
26. In healthcare, because we are so well regulated and we have so many well-established laws and regulations, you can ensure there is always clinical gatekeeping, and the judgment of a healthcare professional is never disintermediated.
27. Our basic approach is therefore to ensure healthcare can be AI-enabled or AI-enhanced, but not AI-decided.
28. That said, we also cannot swing to other extreme, hemming ourselves in with rules and regulations, and failing to harness the opportunities of AI. We must proactively identify use cases for AI in healthcare and adapt our rules and regulations to facilitate them.
29. Part of this effort is to encourage ground-up innovation by our public healthcare clusters. I can see that the National Healthcare Group is doing a lot of it. During my regular visits to hospitals and polyclinics, a key highlight is to witness how such innovations are being done on the ground.
30. For example, at Sengkang General Hospital, doctors are using AI to detect polyps in patients undergoing colonoscopy. The polyps are automatically highlighted in green boxes and appear on the screen. I was there with Dr Koh Poh Koon and he was so excited. It is like having an extra pair of eyes to detect the polyps. This has helped increase detection rates by endoscopists, and also made their job easier as it is less strenuous.
31. At Ng Teng Fong General Hospital, AI tools will analyse the vital signs of warded patients and send predictive warnings to doctors and nurses of possible worsening conditions. I met a nurse who was holding four phones and two walkie-talkies. She said that each one is for a different pilot and one particular phone is for warning her if any patient has a possibility of a worsening condition in the next 12 hours. This AI tool has reduced the number of ICU admissions of warded patients from the wards by over 10%.
32. Doctors at Geylang Polyclinic are using imaging AI to triage chest X-rays. This helps them to prioritise the X-rays with significant abnormalities, allowing patients with more urgent conditions to see a doctor more quickly.
33. To support ground up innovation initiatives, MOH will inject about $200 million over the next five years into the MOH Health Innovation Fund. This is over and above the government’s investments in research and development through the Research, Innovation and Enterprise initiative.
34. However, if we want to make a strong impact with technology, we got to go beyond ground up innovation. Some centralised push will be necessary. MOH will identify proven and impactful technology or use cases, and scale them up into system-wide, national projects. We will start with two projects.
35. First, we will use generative AI to reduce toil. One immediate opportunity is to automate repetitive and time-consuming tasks, such as routine documentation and reviewing of medical records.
36. It is essential for healthcare professionals to do that for every patient, to keep their medical history up to date. But it takes up a lot of time. Further, many patients thought that the doctor or nurse is busy on their computer and is not paying attention to them, when they are actually carefully listening to and updating the record of the patients at the same time.
37. AI tools can now automatically transcribe and summarise conversations between healthcare professionals and patients, and then ingest this information into medical records.
38. The information will still have to be reviewed by a healthcare professional, before becoming official medical records. That is what I meant when I said AI-enabled and not AI-decided. It will therefore not replace the human, but will enhance our efficiency.
39. GovTech has developed a tool to do this, called Scribe, which can handle English, Malay, Chinese, and even Singlish, and it will be progressively rolled out to more public healthcare users. Our health clusters have also adopted various tools to do this.
40. MOH will coordinate the effort of our public health institutions to roll out automated record updating throughout our public healthcare system, before the end of 2025.
41. With that, our doctors, nurses, allied health professionals and medical social workers can spend more time interacting with patients and hopefully, will not be mistaken as being preoccupied with their computer screens and keyboards.
42. Second, use AI for imaging, to improve quality of care. I mentioned earlier how our hospitals are using AI for chest X-rays and colonoscopy. We can make these standard practices across our system.
43. A word of caution here: imaging AI is very powerful and can pick up even the most minute anomalies, but not all anomalies seen in a scan are clinically significant. We should not try to respond to every anomaly detected – it will just raise unnecessary patient anxiety and lead to excessive and unproductive care. We will create a nation of hypochondriacs.
44. There is a phrase to describe this. It was taught to me by Professor Kenneth Mak. It is called V.O.M.I.T., which stands for Victims of Modern Imaging Technology. Hence, we should use AI imaging technology responsibly, to detect and follow up only on clinically significant signs. That way, we can detect these signs earlier, and manage them in good time before they become serious, without causing unnecessary worry and alarm.
45. Other than colonoscopy and chest X-rays, we will also evaluate how AI image analysis can be adopted for screening mammography.
46. If proven effective, from end 2025, we will start progressive rollout of screening using AI for mammography across the system, with the proper workflows and care pathways in place.
47. To accelerate its deployment, we have set up an IT platform, called AimSG. Through AimSG, public hospitals can access different imaging AI models through one single platform and monitor the performance of the models.
Developing Predictive Preventive Care
48. Second, we will use AI to deliver predictive preventive care.
49. There is well known correlation between current risk factors and future diseases. For example, if you smoke or vape, you are more likely to develop many chronic diseases, including cancer or dementia. If you do not exercise and consume too much sugar, you are predisposed to diabetes. If a teenager is deprived of sleep and spends a lot of time on social media, he or she is at higher risk of developing depression.
50. With AI, it is now possible to have disease prediction models that are far more sophisticated and multi-variate, including parameters such as health status, lifestyles, social economic circumstances. The addition of genetic information can make such models even more powerful.
51. In Singapore, about 60 people every day have a heart attack or stroke. With enough data and a well-trained model, it can be possible to pre-warn many of these individuals well ahead of time. We can then prescribe precautionary measures, including changes to their lifestyles or taking some medication.
52. We can therefore avoid pain and suffering in a very significant way. It is a major transformation. We need to be careful with predictions, if not it can also cause a lot of unnecessary worry and alarm. We should therefore proceed deliberately but cautiously.
53. We will start with the first use case, which is Familial Hypercholesterolemia, or FH. FH is caused by defects in our genes that affect the way the body processes cholesterol. Individuals with this condition are more prone to having very high cholesterol levels. They are 20 times more likely to have a heart disease, and at much higher risk of experiencing a heart attack at a much younger age. It is also hereditary, meaning it can be passed down in families.
54. We are starting with FH for this use case, because it is relatively straightforward for us to tell if someone has FH through a genetic test. It is a panel test. The association between the genetic mutations and abnormally high cholesterol levels is also very well established for FH.
55. MOH has been working with Precision Health Research, Singapore (PRECISE) and our hospitals over the last year, to design an appropriate predictive preventive care pathway for FH.
56. Healthier SG family doctors and polyclinics will help identify patients with abnormally high cholesterol levels and recommend them to be tested for the defective gene for FH. The test will be subsidised and the remainder can be paid using MediSave.
57. If an individual tests positive for FH, their parents, siblings and children will be encouraged to undergo the same test. Not the spouse – for obvious reasons – but parents, siblings, and children. This is called cascade screening. By repeating this process, we can pick up as many people in Singapore with the genetic disorder early.
58. We will then advise them to adopt healthier lifestyles as early as possible and start cholesterol lowering therapies, if necessary, to reduce their risk of premature heart diseases.
59. As a matter of fiscal discipline, our health economists at MOH have done the sums. Essentially, we are incurring more costs to do genetic screening and support preventive care. In return, we avoid the costs in treating heart attacks. Our estimation shows that over a 30-year period, we will incur about $14,000 to avoid a heart attack, which is considered cost-effective. This is a pure fiscal calculation.
60. However, in time to come, as the cost of genetic tests continues to fall, we get better at identifying affected individuals, we can test once but use the results to prevent a few diseases, such a predictive preventive care approach may become not just cost-effective, I think it can become cost-saving. This is before counting the avoided pain and suffering.
61. We will start implementing the FH use case from the middle of next year, including setting up a Genetic Assessment Centre. If successful, we will work on other major severe diseases, the possibilities are breast and colon cancers, diabetes, kidney failure, stroke and heart attacks. They will require sophisticated, high-parameter, and multi-variate AI models to be trained.
62. This is predictive preventive care. We will not be able to implement this if we have not invested in genomics research for many years, and we will not be able to do this now if we had not set up the Healthier SG system, which gives us a policy implementation platform.
63. Just think about this – behind the scenes we leverage the wizardry of AI and its predictive powers. At the front end it is a friendly interface, through Healthier SG and through our family doctors.
64. Family doctors will be alerted by the backend systems if their patients are at high risk of specific severe conditions. The doctors would therefore be able to take the necessary medical actions, to counsel and advise the patients on what to do, to avoid a serious disease in the future.
65. With new technology, we enable the family doctor to deliver better care, and empower individuals to better take care of our own health.
Building Up IT Infrastructure
66. Finally, we will build up the necessary IT infrastructure to underpin these new capabilities I talked about.
67. As we use a large amount of data to transform healthcare delivery and prevent severe disease, there will inevitably be concerns about data security and privacy.
68. We had a serious health data breach in 2018. Since then, we have thoroughly reviewed our IT systems and processes, and have been investing heavily in cyber security. For security reasons, I won’t be able to elaborate on what we have done, but let me cite two commonly known initiatives.
69. Number one, we have stored our healthcare data onto dedicated clouds managed by GovTech and Synapxe, where we have full control of the data.
70. Two, our office devices are not fully connected to the Internet. Healthcare staff use a virtual browser to access whitelisted internet sites. Although inconvenient, this is one of the simplest and most effective cybersecurity measure.
71. As for safeguarding data privacy, the government developed the TRUST data exchange two years ago. This is a national platform where datasets can be shared for purposes of research.
72. Through TRUST, patients’ identity and identifiers are removed from various datasets according to government anonymisation standards. Researchers have to submit a detailed request to access data for research and if approved, the requested data will be brought together in the secure TRUST environment to allow analysis. Researchers cannot download any of the data and once the analysis is done, the data is deleted.
73. In the coming years, we will further develop our IT systems to power up innovation, while ensuring cybersecurity and data privacy.
74. The first major development is a consolidated Electronic Medical Record (EMR) system throughout the public healthcare system. Two out of our three health clusters are already using the same EMR system. The final cluster will come on in a few years’ time.
75. By then, I think Singapore will probably be the only country in the world where the entire public healthcare system uses the same EMR System. This will increase the accessibility and quality of data collected at every step of a patient’s journey, to better inform doctors’ decisions across institutions and settings.
76. At the same time, AI tools will be accessible and integrated into the EMR system, to be part of the clinical and operational workflow. It will be embedded and you don’t have to toggle between screens or between software and systems.
77. Second, we will develop the IT platform to train and develop AI tools. As you know, AI capabilities are strictly speaking not developed or built. They emerge after being trained and continuously refined using real-life data. So as you train an AI model, sometimes they surprise us by doing things that we didn’t intend them to do, and their capabilities emerge.
78. This is done through a new platform called HEALIX, which stands for Health Empowerment through Advance Learning and Intelligent eXchange. It will enable the secure sharing of up-to-date, consistent and anonymised clinical, socio-economic, lifestyle, healthcare operations and genomic data.
79. With these data, HEALIX will train various AI and machine learning tools, including for predictive preventive care initiatives I mentioned earlier. HEALIX will be the AI technology factory of the healthcare system.
Closing
80. Today’s speech verges on being esoteric but is probably the most significant I made this year. Because it is many years in the making, it is about the future, it is about a positive transformation of healthcare.
81. I speak before many learned and experienced participants. I hope what I have painted is a sufficiently ambitious plan that is befitting of the technological era that we are living in.
82. If I have, it is due to the work of many experts and agencies, over many years. We have been building this block by block, step by step – from EMR in one cluster, to two clusters, to three, and TRUST to Healthier SG and HEALIX – and now we have a strategic plan for the future.
83. We embarked on this journey because we have been compelled by necessity, for Singapore is ageing very fast and the status quo is unsustainable. Indeed, necessity is the mother of invention.
84. I hope this spirit of innovation and hunger will continue to grow in the healthcare sector and spread to every venture that Singapore embarks on. With this spirit and hunger, I hope we can achieve 15 million years of good life. Thank you very much and I wish you a successful conference.
MIL OSI Asia Pacific News –
