AI in health care

Faster diagnostics, more accurate treatments and better informed patients: just some of the benefits that can be achieved in healthcare with Artificial Intelligence (AI).

For many, however, AI is still far-away. It is something of the future, the investment is too big and the effect too uncertain. AI, however, has long since ceased to be merely an idea of the future. As you read this blog, technology is being used in many ways and especially in the healthcare sector. In order to demonstrate the added value of AI, research agency PwC mapped out the developments and trends.

Consumers accept the operation of algorithms easier when it concerns their private sphere than when it comes to their personal health. Then they indicate that it is fine to use algorithms in healthcare, because who does not also want to benefit from the knowledge of the 1000 best doctors there are in this specialty. Care-giving personnel are more enthusiastic, as long as they stay in charge.

Research agency PwC predicts that the following trends will play a major role in the development of AI in healthcare:

1. Need for a different care system

Ageing, the increase in chronic diseases and rising healthcare costs call for reforms in our healthcare sector. Instead of a ‘hospital system’ – designed for the short-term treatment of patients with acute conditions – we are moving towards a system that enables us to offer long-term care to large groups of people with complex and chronic conditions.

2. Huge increase in medical data

Developments in medicine are happening rapidly. For example, dermatologists by themselves already have around 11,000 new articles at their disposal each year. By 2020, a yottabyte (10 to power 24) of medical data is expected to be available, 80 percent of which will consist of unstructured data. It is an impossible task for people to process this growing amount of data.

3. Changing application of medical technology

In addition to the development of medicine, the application of technology in healthcare is also improving. Whereas technology used to focus purely on medical products (hardware), it is now also widely use for the analysis of medical data. Wearables, big data and now also artificial intelligence and virtual reality are going to provide an enormous improvement in the quality of medicine in the near future.

4. More accessible healthcare

A connection to the internet and mobile devices ensure that we are connected and can collect information 24 hours a day. Patients have much more information at their disposal, can improve their health proactively and can therefore make, together with their caregivers, better decisions about their treatment. As a result, healthcare is becoming increasingly accessible and personal.

5. Growing interest in patients

As a result of developments such as the ‘Internet of Things’, health care is increasingly being adapted to the needs and specificities of the patient. Thanks to smart devices, patients need to come to the hospital less often and can do more themselves, for example measure blood pressure. This movement is also increasingly coming from patients themselves. They are becoming increasingly active with their health and no longer see technology as merely a threshold. This ensures that patients are increasingly open to AI as well.

Applications of AI now and in the future

As I said in the introduction, AI is not merely a concept in the far future. There are already several interesting and promising examples of AI in health care, including: Data management for the better diagnosis of eyes

The Google Deepmind Health project focuses on deep learning. In doing so, algorithms interpret eye scans of patients, teach the system how to identify abnormalities in these scans and advise the doctor on possible subsequent treatments.

It is already the case that a good quality photo of the eyes can diagnose the most common eye diseases better than an ophthalmologist. Hospitals can make use of this technique so that the ophthalmologists themselves have more time for difficult diagnoses and treatments. Ophthalmologists involved are quite enthusiastic (2). Patients react very differently. Some believe in this technique, others trust the doctors themselves more. On average there is a slightly positive reaction (1).

Employees of data companies find it great to work on this kind of projects (3). The hospitals that work with these data companies as customers are also very satisfied (3).

Analysis of the intestinal flora

By analysing the intestinal flora of a large number of people, AI can determine, among other things, whether people are predisposed to type 1 and 2 diabetes. Sometimes a lot of suffering can be prevented by early diagnosis. By combining a precise analysis of this data with data about the diet and blood values of a large number of individuals, it is possible to give spefific personal dietary advice. This often deviates from what has been recommended to diabetes patients until now. It goes without saying that this analysis and personalized advice are highly appreciated by patients (3). Healthcare workers, such as dieticians, are also very enthusiastic (3). They can help their patients much better and see good results.

The holy grail, all the data in one database

Ideally, everyone in healthcare who is collecting and analysing big data related to health would prefer that there is one big database for all healthcare related inforation, such as that collected by hospitals, general practitioners and other healthcare professionals. This vast amount of data would be a haven for artificial intelligence and could provide significant insights for both healthcare development and highly personalised advice to individual clients. There are numerous practical objections, ethical issues, safety risks and conflicts of interest that can potentially stand in the way here. However, if such a database could be realised, it would be greatly appreciated by both consumers (3) and healthcare professionals (3).

Issues to consider when applying AI in the healthcare sector

The applications of AI are promising, but I would like to make a few important remarks. In order for AI to be carefully integrated into our healthcare system, it is important that ethical standards are established and that such developments of AI takes place only gradually. In this way, the possible downfalls of AI can be anticipated in time and patients can gradually become familiar with them. Additionally, it is important that healthcare professionals acquire basic knowledge about AI in a medical environment and that AI-developing companies (continue to) communicate well about the benefits and risks of AI in healthcare. Finally, the quality of the data used determines the success of the deployment of artificial intelligence. After all, ‘bad input’ denotes ‘bad output’.

In this way, the applications of AI can actually respond to the trends in healthcare and doctors and patients can make use of the technological possibilities.

It is especially important who will manage the data and take care of the security. Will this be the government? A conglomerate of healthcare providers? Or even a commercial party that will collect and analyze the data. Research shows that clients are very suspicious of management by commercial parties (-3). Employees at data companies also see no point (-3) to develop algorithms for large commercial parties. They prefer to do this in consultation with the healthcare providers.