Skip to 0 minutes and 10 secondsThen I will go to the most important topic of this class, that is participatory healthcare. Basically what I mean as I explained it's about how we empower patient by providing him more information about his own healthcare to participate in management of his or her own healthcare issues. There is one good example in Taiwan actually. Through government of India it's called my health bank, where government open data for the patient's to see their own health records they own. So they can't see others health record or information. So I need to put my own card and my own keywords.

Skip to 0 minutes and 53 secondsSo there are very strict, stringent security distinct layer, once you pass through this, you have assessed to your own medical electronic health records which are from National Health Insurance Bureau. And this is the first page you can just go and try or you can try to google. People from Taiwan can actually use and see how it's working. So as you see in the screen you can key in your, what to say, insurance card number and other these things, then you can log in, and you can practically see how many times you have visited the hospital. Not only hospital, dental clinic or traditional medicine, wherever you go, you will have those information there. And also you can add information.

Skip to 1 minute and 55 secondsSo for example, if you have some appointments, if you have some dates there, you need to go and collect your medications refilling, or go for vaccinations, or visit clinic, you can add this information in your diary of your my health bank. And the most important thing is it's not only you see the information from the EHR or medical record but also you can add your own physiological data. You can add your other vaccinations, or habits how you are sleeping, you can add your blood pressure, you can add how are you are physically active in a given day. So it's sort of a diary where you can input information.

Skip to 2 minutes and 42 secondsAnd another important thing is it also gives you prediction of some serious diseases. That is the key of my health bank. And I think this is how we are encouraging people to come back and see what's happening with them from time to time, and see whether they are having any high risk of getting any serious diseases. So for now the my health bank is using, what to say, chronic kidney disease, and hepatocellular carcinoma, lung cancer, and kidney disease, as prediction.

Skip to 3 minutes and 25 secondsSo once you go there, the system has your all comorbidities, the system has all medications you are taking, the system also has your lab values, and it analyzed all those things, and use AI for analyzing this data, and it predicts in next 10 years are you going to develop what are the chances or what are the risk or for you to develop either kidney disease or lung cancer. But the issue is it's so long time, ten years. So we have developed a steady where we can predict your cancer lung cancer in three years and one year. Because I think that is more appropriate to take some real action.

Skip to 4 minutes and 13 secondsSo if you say you will going to develop in next ten years, in next 20 years, you really can't take any initiative to avoid that. But if I say it's going to happen within three years or within one year, you will be more proactive, you will be more ,what to say, motivated to change your behavior. Because changing behavior is the key for maintaining good health life or healthy life. So you need to change your behavior, and see what is good, what fits for your body, and what fits in your situation. And as you see here, you can put your blood pressures, and puts all those information day-to-day in your health bank.

Skip to 5 minutes and 4 secondsOnce you go back to your clinic, once in a month, or once in a two months, then doctor can just see a graph of things what's happening when you are at home.

Skip to 5 minutes and 16 secondsAnd this is the study I was discussing which we have done. Before going to study, I want to do one ,what to say, outlook a blue map of our data, medical data of our body. So once we get birth, three things comes with us. One is genomic information because that is given by our parents. And comes the exposons or environment because it totally depends where you are born. You born in which country which city, or how industrial area or it is, what to say, in a natural place or it is industrial or what type of facilities are there.

Skip to 6 minutes and 8 secondsSo these are the just given, we don't choose our place of birth or, what to say, location of birth and environment of our birth. And but the third thing is behavior, here comes the key. So we can change our behavior, but we can't change our genomic information, we can't most it's not so easy to change our exposons or expose types, but we can change our behavior. We can not smoke, we don't need to drink, we can eat one day healthy food, we can do physical exercise if we want . Otherwise, just you can do what is good for you. After saying these three things, once you go to hospital, then you get the things which are inside the circle.

Skip to 6 minutes and 59 secondsLike you are demographic, like your diagnosis, your medications, like your lab values. So, you see all these type of data is collected, once you start going to hospital or get into, what to say, health care management. And here comes what we can do using this data. We can predict, we can early detect, and we can give recommendations. So treatment rehabilitations are the things of afterwards months. Once you get disease there is nothing else we can do. Definitely, we need to treat. After treatment, we need to rehabilitate. But the most important thing is prediction and early detection. So that you suffer less and it's good for everyone. Then as I told, we started predicting about lung cancer.

Skip to 7 minutes and 57 secondsBecause the same purpose of we want to change the behavior. We want to change the behavior of the people. And as I told, 10 years is too long and one month is too short. That's why we have chosen one year to three years span to predict lung cancer.

Skip to 8 minutes and 21 secondsThis is how it works if you view images different type of animals, and you tag them okay, this is cat, this is dog, this is cat, then once you show another image, the testing image, then the system will see extract the futures or patterns of that image And it can successfully say whether it is cat or a dog. So same thing happens with our medical data once we put all our values there. So this all shows you details of how convolution neural network works. So it works with the image, image patterns, and with the pixels, it calculate and it, what to say, converge different pixels from the neighborhood, and come up with the results.

Skip to 9 minutes and 18 secondsSo actually this is how the computer will be seeing things, natural things. Like if you put the images of dogs, you see how the computer see, and do the answer. So these are on left side are the images, on other side are the analysis done by the computer. Similarly, we did for our study, what we did is we put all medications, we put all, what to say, comorbidities diagnosis in the system, and then see people using these these drugs, and people having these these diagnosis.

Skip to 9 minutes and 59 secondsWe call it ICD, International classification of diseases, who are developing, what to say, HCC, hepatocellular carcinoma, and we can also predict 3 years ahead and 1 year ahead And this is how the current system is showing in my health bank. Ok. For you, you are from male, and you are in this age group, so they give points for each, and then you say you have risk of this many to develop, what to say, hepatocellular carcinoma in next 10 years.

Skip to 10 minutes and 39 secondsBut in our system we knew them in short time, in next 1 year or 3 years So this is the data which we have used, and these are the icd-9 CMS, and these are the long-term medication we used, because we don't want to use with short term medications, because that could be so many, so we use long term once we use. This maybe they can have some adverse effect, and then we see who are prone to develop hepatocellular carcinoma. So this is how it works and this is exactly what I was talking about.

Skip to 11 minutes and 18 secondsYou put inner layer, then there you will be having, no sir, you put input variables, then you have layers, and you come up with the two output variables, like outcomes, our positive of cancer or no cancer. And here you see for one year we got up to 91%, and for three years, we got up to 88% of accuracy area under the curve. For those who are not familiar with a AUROC, it's area under the ROC curve, and which use you the accuracy sensitivity and specificity of the model. So it means it's 91% time we are accurately predicting if we say you get cancer, you will get the cancer.

Skip to 12 minutes and 14 secondsAnd this is how again explaining the results of three years and one year. This can be the potential output. So once I know, okay, I have these these disease, I was taking these medications, so what I should do? Maybe I should not drink more alcohol. Maybe I should take care of the medications which will not affect my liver function. And I should, what to say, avoid fatty food. So these are the, what to say, recommendation we will be providing of we do all this analysis.

Participatory healthcare

Dr. Shabbir explains how we empower patients by providing information for personal management in healthcare. An example of my health bank is introduced in the session. We will discuss how AI can improve personal disease prevention and prediction.

Do you have anything like My Health Bank where you are from? Can you easily access your own health records? What might be the good and bad aspects to seeing a prediction of your future health in this way? Feel free to leave some comments below.

Share this video:

This video is from the free online course:

Artificial Intelligence for Healthcare: Opportunities and Challenges

Taipei Medical University