The healthcare industry is currently experiencing a revolutionary transformation, shaped primarily by two factors: the COVID-19 pandemic and technological innovation. While the pandemic clearly necessitated quick and modern solutions, the existence of technologies like artificial intelligence (AI) and machine learning (ML) provided a much-needed boost not only for COVID-19 response, but for treating other illnesses as well.
As per the old adage, “Prevention is better than cure,” and rightly so, because not only are medicines and treatment getting more and more expensive, especially for chronic illnesses, but new ways of disease prevention and management are coming to the forefront— thanks to the wonders of technology.
During the most recent Healthcare Frontiers online conference, Jicara Media gathered medical and IT experts to discuss how the latest technologies are providing a shot of adrenaline to healthcare transformation across the globe.
Among the specialists there was Dr. Praveen Deorani, Senior Data Scientist for the Ministry of Health Office for Healthcare Transformation in Singapore, who presented a case study titled “Machine Learning Models for Patient Education and Activation”. During his presentation, Deorani argued that the quality of medical care is the least significant determinant of all health outcomes.
“Behaviour and lifestyle are the most significant factors, but they are also the most actionable,” Deorani said.
Deorani, who is advocating for a more proactive role among patients in terms of their health and well-being, stressed the importance of taking ownership of health maintenance.
“The chronic diseases will stay with the patient even when they are at home, even outside the healthcare setting. That’s important because, for example, a newly diagnosed diabetic might spend only 10 minutes every six months in the clinic, and the rest of the time they spend at home or in their personal life, but the chronic disease stays with them even at that point,” Deorani noted.
“Lifetime behaviour is not only greater at the onset of disease, but also (a) predictor of the progression of the disease. One of the important things there is compliance to treatment, or medication adherence. But you know, there is evidence that consistently shows that about a large number of patients do not take their medicines regularly,” he added.
The harsh reality faced by doctors when it comes to patient management, is that most patients have more than one medical condition. So it’s not just plain diabetes, for example. The same diabetic person might also have hypertension, or a heart condition, which creates a barrier for risk communication.
“In reality, most of the patients who are at risk of (the) onset of disease do not have any one factor which is out of control. Usually, they have a lot of risk factors,” clarified Deorani. “That’s what happens with most of the patients. That’s why it’s hard for patients to understand exactly how much risk is there in the progression of disease. The other thing is the understanding of the risk factors, because the risk factors for (the) onset of disease are sometimes different from the prevalence of the same disease.”
“That overwhelms the patients because there are multiple things they must do. They must lose weight, they must reduce their smoking habits, they must control their cholesterol, they must exercise, and so on. Most of the early stage chronic disease patients know all these things. And yet it’s very hard because there are so many things that need to be taken care of. And it overwhelms them because changing their lifestyle significantly, very quickly is hard,” he added.
As early as now, significant investments must be made towards patient education and activation, otherwise countries like Singapore will be spending a lot more on healthcare.
“It’s estimated that the healthcare expenditure in Singapore will increase by more than 100% in the next 10 years. And so, if the role of patient management is the key to address the challenge, how do we go about that? How do we educate the patients? How do we empower the patients? How do we affect a sustained change in their behaviour?” Deorani asked.
For the IT health expert, it was clear that technology could be of some assistance to achieve this end.
“Most of the patients who are at risk have multiple borderline risk factors. It’s not obviously clear sometimes, which is the one intervention they should start. So that brings us to (the) possibility of a patient education tool. You could think of this as a tool that a clinician could use, or a patient educator could use during the counselling visit with the patient,” Deorani said.
The power of ML
In the same way that ML minimised risks in industries like energy, manufacturing, and even agriculture, the technology is eyed as a game-changer in patient management, by leveraging data to reduce morbidity and mortality.
Deorani detailed how their patient education tool, which will soon be used on actual patients, works.
“It’s a neural network (that) is typed over four different variants. What it does is it then predicts the risk trajectory of the disease— chronic kidney disease, chronic heart disease, congestive heart failure, and cerebral vascular disease. The key thing here is that this same model predicts multiple diseases, and the trajectory of multiple diseases, which is quite different from most of the models that you see in literature. Typically, they focus only on one outcome or one start point, such as, for example, these models may talk about progress, from diabetes to kidney disease, or kidney disease to death, and so on. But typically, they don’t focus on multiple diseases, which is not right, because as I said, most of the patients (have) multiple diseases to begin with, and multiple risk factors. We need to understand the holistic picture,” he said.
The tool, however, is limited to patient education, and while it will enable patients to better understand their health and well-being, it cannot assure 100% patient adherence to their prescribed medical regimen.
“We can see that nowadays having knowledge is not the same as behaviour change, and there (are) a lot of research (that) consistently shows that the patient education programs, and the results are always very similar. It shows that patients after six months, they know very well about their medications, they know the importance of lifestyle, and yet there isn’t much change in their behaviour,” Deorani noted.
“People learn differently. Some people may prefer videos, some people may prefer articles, whereas some may prefer just a short bullet— just a short, actionable item. They don’t need to know the biology; they just want (to know) ‘What should I do?’ We need to understand these aspects, these different parts of a patient’s personality, as well as the clinical risk profile. And we need to tune our methods into them,” he added.
Detection is still key
The power of data to predict future medical trends has to be leveraged more, Deorani emphasised. One example, he noted, was their telehealth services, which makes use of data to improve patient care.
“We have patients at their home, they have these wearable devices, blood pressure machines, (a) Fitbit, and so on. And they mess up themselves at home. This data is then captured through their mobile phone, and it’s transmitted into a back-end server, where the data is analysed and understood what’s going on. What that means is if there is an event of clinical significance, such as their blood pressure is too high for a significant amount of time, or if their heart rate is too low, or so on—if there is a trend month over month that is increasing, even though they are on medication, something is wrong. Maybe the doses need to change. Those types of events are detected using this data analysis and they automatically go to the dashboard, (then) the physician can contact them,” he explained.
“The goal now is to build on this existing knowledge and exchange research to develop this personalised tool for further education and encouragement and innovation,” Deorani concluded.