12-12-2022 | By Robin Mitchell
As the capabilities of AI continue to improve, the medical industry is still very reluctant to utilise AI, but this may no longer be the case now that Google AI algorithms are being integrated into mammogram machinery. Why does AI face resistance in the medical field, how will AI be used to help detect breast cancer, and are we about to see the mass use of AI in medical diagnostics?
Why does AI face resistance in the medical field?
Anyone following my work on Electropages will be fully aware of my hatred for the inefficiencies plagued by the NHS, the poor response times from GPs, and the lack of centralised systems that can correlate data across numerous departments. At the same, I also find the resistance that AI faces in the medical field by doctors truly shocking, which raises the question of whether this resistance stems from the fear of losing one’s job.
But while there are many conspiracy theories regarding the true nature of AI, who is developing it, and how it is being used, there are some very tangible reasons why integrating AI into medical diagnostics is problematic. One of these reasons is that, unlike humans, AI cannot determine if a patient is lying, what their current emotional state is, and other subtle hints that may suggest different diagnostics. However, many GPs are very much like robots themselves who often try to get through patients as fast as possible and will often turn to gpedia which itself is just a Wikipedia for conditions. This problem is made worse when considering that the NHS is a lottery system, with some doctors being superb and others being abysmal (I am very fortunate that my local GP does an excellent job).
Another valid reason why AI faces challenges in medical applications is the need for vast quantities of medical data. Just like bank accounts and national identity numbers, medical history is as personal as data can get, and handing over data without patient permission is illegal in every sense possible. Patients could be presented with the opportunity to share their medical history with AI developers, but this could lead to confusion and misleading practices, especially if commercial interests are involved.
Google AI software to be used in mammograms
Recently, a medical technology company that produces solutions for breast cancer detection, iCAD, has announced that it will integrate Google AI solutions to help medical professionals improve detection rates. Earlier research into AI technologies and breast cancer detection published in nature showed better detection rates and smaller false positives and negatives when using AI. With over 90,000 mammograms to train from, the resulting AI was able to lower the false positive rate by 6%.
To help radiologists with diagnosis, the AI is entirely cloud-based, which reduces the computational burden on those who lack resources. At the same time, the use of cloud-based AI also opens up the platform to all iCAD customers across the globe, which could eventually lead to low-cost diagnostics capabilities. Furthermore, increased use of AI will only improve its ability to identify cancerous tissue, eventually leading to an AI that can effectively eliminate the need for human intervention.
Are we about to see mass integration of AI technologies in medical sciences?
The use of AI to improve breast cancer detection rates using cloud-based software is a significant step for AI in medical applications. Not only will the software help to catch cancers early (when they are straightforward to treat), but its eventual success will demonstrate to the medical world that AI in diagnostics is the future.
Of course, we can expect to see large amounts of resistance to AI over the next few years, and the opposing side will likely weaponise human faces and privacy. For example, there will probably be doctors stating that “AIs cannot read human emotion” and “all your medical history will be accessible by engineers” in an attempt to scare the public.
It is unlikely that AI will ever replace humans entirely, but if paired with a human, they can be extremely powerful. In the future, the most advanced GPs will utilise AI to try and look for unusual symptoms and links, while the GP will judge the performance of the AI, read the emotional state of the individual walking into the room, and make a joint decision with the AI.