An all-encompassing environment is needed first.
Historically, the expert system MYCIN, developed in 1974, is considered as being the very first application of artificial intelligence in healthcare. Its purpose was to assist doctors in their diagnosis and in the treatment of various blood diseases. By attempting to reproduce expert cognitive mechanisms in a given area through reasoning based on facts and rules (around 700 rules), this software has led to more conclusive results than human attempts. In 1979, in order to gauge its effectiveness, the system participated in a competition against 8 physicians: 10 cases were submitted – and MYCIN ranked first every time.
The promises of AI’s application to the health sector are multiple: as a result of better diagnostic accuracy in particular, it would reduce operational costs for medical institutions and lead to increasingly personalized treatment plans for patients.
Fantasies are commonplace when it comes to AI. However, it is necessary to find the right environment for implementation in order to fully embed AI within the healthcare field.
Interoperability paves the way for precision medicine
Interoperability, referring to the ability to easily share data between different stakeholders, is an inherent constraint within the health field, hindering the integration of AI. Central medical databases or standard storage formats are very rare – most of the time documents are handwritten and then faxed, making any kind of data extraction not interpretable for machines.
In order to establish a fostering environment for AI in healthcare, it is paramount to democratize access to medical data. This is of utmost importance to feed and improve machine learning models.
Multiple factors can have an influence on it, such as:
- The volume of medical data increase as a result of either the multiplication of wearables or genomic sequencing databases.
- Data availability facilitated by the widespread use of electronic health records. They now allow easy access to a data history going back more than 10 years.
The French start-up Lifen, for example, is now primarily focused on interoperability in the health sector. It has developed a secure exchange platform for physicians and medical institutions, powered by artificial intelligence algorithms. In addition to providing them with the most complete directory of healthcare professionals in France, their solution is compatible with 100% of medical practice software: the data format used no longer hinders the sharing of medical data, and patient information is no longer scattered across multiple softwares.
The advent of telemedicine is a perfect example of the democratization of medical data access.
Nowadays, the mobile phone provides a new source of data and represents a powerful diagnostic tool. The ever-increasing use of mobile phones combined with reduced visual recognition algorithms error rates have prompted the founders of SkinVision to design an application using the phone’s camera to monitor skin lesions and assess skin cancer risks. As a result, this “telemedicine” approach allows doctors to build up a more refined knowledge of patient’s behaviours and lifestyles, and to develop medical treatment plans that are more preventive than curative.
Due to the combination of Big Data, predictive analysis and artificial intelligence, diagnosis and treatments are personalized, leading to an utterly client-centric approach – nowadays referred to as Precision Medicine.
Regulation flexibility and government support are both needed in order to quickly implement AI applications within the health field
The frantic pace of machine learning and deep learning systems upgrades set the tone and a loosened regulation is needed for AI implementations in healthcare to keep up.
In the United States for example, the FDA (Food and Drug Administration) has been fast-tracking the approval process of AI diagnosis software and the use of computer vision in medical imaging since 2018. In April 2018, the organization has cleared AI algorithms to diagnose eye disease such as diabetic retinopathy without the second opinion of a doctor.
Similarly, major tech companies in China are strongly supported by the government, particularly in the centralization of data and the automation of medical processes. China’s Ministry of Science of Technology announced in 2017 that they would be relying on Tencent to launch an open source AI platform for medical imaging and diagnostics. In the meantime, the use of Tencent’s network platform, WeChat, to book online appointments and pay fees is increasingly spreading within medical facilities. The strategy seems to pay off as in the first half year of 2018, China overtook the United Kingdom to become the second most active country in applying AI within the health sector.
Developing synergies between tech companies and medical institutions is a prerequisite for the integration of AI within the health sector
Tech giants are facilitating data access and providing continuously optimized technological tools for medical facilities.
Apple for example contributes by building an ecosystem conducive to clinical trials. Processes around clinical trials can turn out to be particularly long and costly. By providing researchers with a continuous data flow from portable devices such as the Apple watch, researchers can benefit from real-time information on the health of patients participating in their studies. The American tech giant has also launched two open source frameworks in 2015, ResearchKit et CareKit, facilitating the optimal match between patient and clinical trial, as well as helping to monitor their progress during the initiated study. This initiative is just one example among dozens of others undertaken by Apple since 2015, as a result of the company’s desire to position itself as a leader in the sector.
Thus, interoperability, government support and flexible regulation, as well as the introduction of synergies among tech companies and medical institutions, represent the interlinked environmental variables that can provide the fertile ground necessary for the full exploitation of AI’s potential within the health industry.