While the aims and vision of healthcare have remained consistent since time immemorial, the healthcare landscape and care delivery have been transformed by technology. New technologies are continuously emerging and there are a plethora of opportunities to utilise them to benefit our lives. However, as with all innovations, the impact of technology on healthcare has its blessings and pitfalls. Here, we focus on how Artificial Intelligence (AI) is transforming the healthcare industry.
PREDICTIVE ANALYTICS & AI
Introducing an innovative form of AI, Predictive Analytics. Lately, healthcare stakeholders have been making use of this relatively new form of preventive medicine for a number of reasons. Relying on the vast improvements in computational power, predictive analytics has the ability to synthesise insights and trends from masses of data. This is done through data mining and modelling with the use of statistics.
Modelling in AI is a process whereby a human’s decision-making process is replicated in machines. Simple or complicated thought processes can be encoded in instruments or devices. An example of this is General Electric (GE)’s invention, Digital Twin. By utilising predictive analytics, GE has been able to create an orthogonal method of predicting the health and performance of its assets over their lifetime. In the aerospace industry, Digital Twin is able to analyse masses of manufacturing, inspection and repair data to predict failure modes of jet engines. This is an extremely useful way to prevent catastrophic events, such as mid-flight engine failure, and allows for optimisation of services and maintenance.
By structuring and ingesting medical parameters, predictive analytics algorithms are able to evaluate and analyse how these biomarkers can potentially affect individuals’ health trajectories and provide prognostic diagnoses.
Identifying equipment maintenance needs before they arise
Healthcare operations can benefit from prognostics. Certain components of medical equipment such as MRI scanners degrade over time through regular use. If one is able to predict when a component needs replacing, one can minimise unscheduled workflow disruptions that hinder both care providers and patients. Photo: iStock
Predictive analytics can also be used as additional support in the healthcare industry. Advances in technology have given rise to an abundance of underutilised medical data stashed in silos. This data could have been obtained from diagnostic tests, imaging, genomics and many other new medical technologies. Predictive analytics can effectively put this vast amount of medical data to good use. By structuring and ingesting medical parameters, predictive analytics algorithms are able to evaluate and analyse how these biomarkers can potentially affect individuals’ health trajectories and provide prognostic diagnoses. With more information at our disposal, algorithms can lead to better quality decisions to assist healthcare professionals.
A notable example is how Mayo Clinic, a world leading health system based in the US, is utilising predictive analytics to enhance patient care. According to the Commonwealth Fund’s Promise and Peril 2021 report, Mayo Clinic has developed an algorithm that can “unmask heart abnormalities long before patients begin experiencing symptoms”. 1 This revolutionary breakthrough makes it possible for healthcare professionals to prevent the onset of stroke and to predict arrhythmia in the future. Instead of reactively diagnosing individuals when cardiovascular-related diseases occur, it is now possible to foresee disease progression, giving individuals time to adjust their lifestyles or start treatment. The Mayo Clinic algorithm is in its trial phase but has the aptitude to show promising results.
AI and data analytics in Singapore hosptials
In the light of COVID-19 outbreaks, the Integrated Health Information Systems (IHiS) team in Singapore swiftly made an enhancement to their existing Command, Control and Communications (C3) System at Tan Tock Seng Hospital for better pandemic management. The system helps healthcare workers to predict demand for hospital beds and other healthcare resources, identify bottlenecks such as long waiting times for patients and or at laboratories performing a large number of patient tests, as well as safely manage the crowd flow as COVID-19 numbers surged. Source: Centre for Healthcare Innovation
In addition to its potential applications in preventive medicine, predictive analytics can be utilised in various other settings in healthcare. It can be a useful tool for government organisations to monitor population health demographics and ensure future needs are taken care of. In Australia, the Department of Health and Human Services (DHHS) is able to work on improving their health services to the Victorian community with the help of predictive analytics. By studying data and trends, DHHS was able to get more information regarding:
- Supply and demand modelling of the health workforce;
- A hospital inpatient protection model; and
- An emergency department projection model.
These projections are useful for resource allocation and the planning of future healthcare needs. The opportunities presented by predictive analytics are endless.
WHY NOW? CURRENT PAIN POINTS?
The rise in the use of predictive analytics can be attributed to a few factors. Firstly, it is widely known that the world faces an ageing population and rising healthcare costs, both in the East and the West. The steady increase in patient numbers and the eventual strain this puts on the healthcare workforce presents an urgent case for the use of predictive analytics to optimise clinical workflow efficiency. Since much time is spent on administrative tasks, automation could greatly reduce the time spent on these less significant tasks and time savings can be redirected to patient interaction. Predictive analytics structures medical data in a comprehensive way for doctors to present to patients, and allows them to focus on their patients. By optimising staffing requirements and scheduling, as described by Hewlett Packard Enterprise (HPE), healthcare costs can be reduced as well.2
DARA (“Disease-Associated Risk Assessment”)
Among patients, there seems to be a lack of understanding about the value of health screening procedures, as reported in a study conducted by Chien et al. (2020), which lists this as one of the top three reasons why patients do not want to go for health screening. Another complaint patients seem to have is the long waiting time in clinics. These reasons contribute to suboptimal care delivery. However, these issues can be overcome with the use of predictive analytics to streamline clinical workflows to create more efficient systems and present health data in a clear, comprehensive format.
To address these issues, Mesh Bio has developed DARA, which seamlessly integrates predictive analytics and health screening. DARA is an engine that draws in all diverse types of multi-dimensional health data in a meaningful contextual way to automate the structuring and synthesis of health insights. DARA takes into account patients’ demographics, medical history, biomarkers, amongst others, to generate actionable insights which use specific goals to help patients take control of their health. This clinical decision support reduces uncertainty in care delivery and increases the effectiveness of interventions, thus improving patient outcomes and reducing chronic disease risk.
Additionally, this manpower strain in the healthcare system and workforce has been further exacerbated by the COVID-19 pandemic. With an overwhelming number of COVID-positive patients, healthcare professionals are now busier than ever. With the ability to triage, predictive analytics can aid doctors, nurses and management teams in knowing where to direct their time and expertise. This alleviates service failure from overstretched healthcare resources and allows healthcare stakeholders to deliver and maintain quality care delivery.
Additionally, there has been a surge in the number of metabolic diseases worldwide. According to the World Health Organization (WHO), there are 17.9 million deaths annually due to cardiovascular diseases.3 The Lancet also observed that approximately $825 billion (in international dollars) is spent on diabetes care each year.4 These are worrying trends for both developed and developing nations. These metabolic diseases, such as diabetes, hypertension and cardiovascular disease, typically arise silently and only present noticeable symptoms at advanced stages or during acute catastrophic events such as heart attack. These diseases can be attributed to genetics, but mostly due to one’s lifestyle habits or medical history. Predictive analytics offers an opportunity for individuals to understand how their lifestyle choices are likely to impact their health future. By analysing their medical data, one is able to provide individuals with a better understanding of their own health.
Legal framework in telemedicine
While some countries, including the UK and France, view e-hospitals as a possible alternative to brick-and-mortar ones, others, including Germany, Russia and China, regard them as supplementary health care services. China specifically prohibits initial diagnosis or treatment via the internet, whilst allowing virtual follow-up visitations and online prescriptions. Photo: iStock
This is especially important during the pandemic now. It has been proven that individuals with underlying chronic diseases have a higher risk of developing severe COVID-19. The Centers for Disease Control and Prevention (CDC) in the US has characterised chronic diseases as heart disease, diabetes, cancer, chronic obstructive pulmonary disease, chronic kidney disease and obesity.5 Now more than ever, healthcare professionals are focused on tackling metabolic diseases to prevent catastrophic effects from COVID-19 with the help of predictive analytics.
Now, the introduction of predictive analytics is yet another big step towards bridging the gap between patient and doctor.
READINESS OF THE HEALTHCARE INDUSTRY FOR PREDICTIVE ANALYTICS
The healthcare industry has advanced tremendously in the past several decades. At the most basic level, the keeping of physical medical records is increasingly being replaced with the global adoption of the electronic medical records system (EMRS). Automation and natural language processing technologies have greatly increased the efficiency of medical note-taking and transcribing. Cloud-based storage services have been implemented to allow interoperability. Now, the introduction of predictive analytics is yet another big step towards bridging the gap between patient and doctor.
Patient-owned health data system
In Estonia, 95% of health data is digitalised. The country has an open source health information and patient-owned data system. People have unlimited access to their medical history, referrals, prescriptions, dental records, health insurance and consent forms. They can choose what information they want to make completely private and inaccessible to providers. Photo: Elisabeth Blanchet / Alamy Stock Photo
The COVID-19 pandemic has catalysed the use of telehealth and accelerated its widespread rapid adoption. NEJM Catalyst describes “telehealth” as a mode to deliver health care, health education, and health information services via remote technologies.6 The risks presented in in-person meetups have caused clinicians to shift toward providing health services digitally. Healthcare professionals, who do not physically meet their patients, have to rely on data to monitor trends and treatments and to aid in diagnosis and its results. Tools such as MedWhat and Fitbit can augment the work of healthcare professionals and enhance the patient care journey.
RECOGNISING AND AVOIDING PITFALLS FOR AI IN HEALTHCARE
Despite its promise, the use of AI and predictive analytics in healthcare is not without limitations. As discussed by HealthITAnalytics, data privacy and security can be a huge issue when it comes to digital medical information.7 Thus, cybersecurity should be every healthcare stakeholder’s priority, to protect the interest of their patients. By implementing strict protocols and following through, healthcare companies can ensure that their patients’ confidentiality is maintained.
There are those who are concerned about the use of AI in healthcare due to potential biases that may stem from inaccurate datasets or inadequate representation. Since algorithms take in numerical data or trends, it is possible that social factors and influences are left out when using predictive analytics. However, this can be overcome if researchers and creators of the algorithms are mindful of the possibility of biases. As discussed by the Healthcare Information and Management Systems Society (HIMSS), “effective datasets must be based on a diverse range of race, gender or geography in order to avoid biased algorithms.”8 On top of that, users of AI must adopt impartial practices and effectuate guidelines to prevent the introduction of biases.9
By implementing strict protocols and following through, healthcare companies can ensure that their patients’ confidentiality is maintained.
CONCLUSION
Beyond what has been discussed above, healthcare professionals will always prioritise their patients first. Countless promising inventions have revolutionised healthcare and it should be no surprise that there are more to come. Whilst addressing the issues faced by the healthcare sector, we must bear in mind the fundamentals and potential perils. The human brain, and now AI, make anything possible.
A helping hand with healthcare costs
By using EMMA to do the labour intensive massages, physicans can now offer a longer therapy session for patients while reducing the cost of treatment. But the robot will not be replacing its human counterparts just yet. Source: www.alphr.com
EMMA, the Robot Therapist
Tuina (“推拿“), a form of traditional Chinese therapeutic massage, has been used in China for centuries as a treatment for various conditions, including joint injuries, muscle sprains and nerve pain. Though only more recently introduced in the West, the use of Tuina has been gaining momentum as a form of ‘alternative’ physiotherapy treatment offered by some hospitals worldwide.
Unlike normal massage, Tuina targets specific acupoints on the body. Tuina practitioners brush, knead, roll and apply pressure on and around these acupoints to reduce blockages in the body’s circulatory system and improve its flow. Good circulation reduces pain and the risks of illness. Each Tuina session is very much a combination of skilled know-how with a considerable amount of manual manipulation by a trained practitioner.
Enter EMMA, or “Expert Manipulative Massage Automation”, developed by Singapore company AiTreat Pte. Ltd., to lend a hand, literally. EMMA is a robot masseuse designed to help manage the heavy reliance on qualified therapists. “In fact, EMMA is a cobot – collaborative robot. It works alongside a human, in this case a TCM practitioner, and takes over about 80% of the repetitive and labour-intensive work, leaving
the remaining 20% to a human expert,” says Dr Albert Zhang, the designer of EMMA and founder of AiTreat. “That way, it solves the labour shortage problem in therapeutic healthcare and helps bring down treatment costs.” Lower cost and improved accessibility to therapeutic healthcare is especially desirable in communities with an ageing population.
EMMA is equipped with sensors to measure muscle stiffness and uses 3D vision technology to analyse each patient’s body. This data allows patients to better understand their own conditions and enables physicians to make more precise judgements for diagnoses. Artificial Intelligence is used to determine the best course of action, and EMMA’s robotic arm performs the therapeutic massage.
Unlike therapy by human massage therapists which are sometimes prone to inconsistencies (particularly for longer-term treatment plans), EMMA maintains diagnostic and treatment records for individual patients, and guarantees unchanging precision and effort, even between sessions. Patients have control over EMMA’s features during the massage treatment: they can adjust the massage strength, temperature at the touchpoint, and are able to halt the procedure at any time. During the COVID-19 pandemic, the use of a robotic masseuse also presents a good alternative for patients who might wish to minimise human contact.
According to Dr Zhang, the upcoming version of EMMA will also have the ability to perform western physiotherapy treatments, for example, shockwave therapy and ultrasound therapy. With that, EMMA will become a highly versatile healthcare cobot equipped with the best of both Eastern and Western worlds.
DR ANDREW WU
Dr Andrew Wu is the Co-Founder and CEO of Mesh Bio, a digital health startup providing automation and analytics software solutions built on AI, for the transformation of care delivery. He was previously Chief Operating Officer of Clearbridge Biomedics (now known as Biolidics) where he developed innovative cancer diagnostics technology in the field of liquid biopsy. He was also Chief Product Officer of Clearbridge Health. Both companies are now listed on SGX Catalist. Andrew holds Bachelor’s and Ph.D. degrees in Biochemical Engineering from University College London.
DECEMBER 2021 | ISSUE 9
Tomorrow's Technology Today
https://www.statnews.com/wp-content/uploads/2021/04/STAT_ Promise_and_Peril_2021_Report.pdf
- https://www.hpe.com/us/en/insights/articles/5-ways-to-cut-healthcare-costs-with-predictive-analytics-1707.html
- https://www.who.int/health-topics/cardiovascular-diseases#tab=tab_1
- https://www.thelancet.com/journals/lancet/article/PIIS0140-6736(16)00618-8/fulltext
- https://www.cdc.gov/pcd/issues/2021/21_0086.htm
- https://catalyst.nejm.org/doi/full/10.1056/CAT.18.0268
- https://healthitanalytics.com/news/arguing-the-pros-and-cons-of-artificial-intelligence-in-healthcare
- https://www.himss.org/resources/uncovering-and-removing-data-bias-healthcare
- https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7277138/
