Looking back, work on AI has continued for a long time, and health care has always been a part of it. Back in the decade of the 50s and 60s', it was all about creating databases, e.g. electronic medical records and clinical information data. Such works eventually led to the launch of search engines like PubMed. Similar biomedical search engines saw a lot of growth at that time, and Ivy League universities started to use those more and more. This created a platform for enhanced collaboration among researchers.
The use of AI in modern healthcare is not that uncommon. Online scheduling of appointments is an example. Evidence-based medicine, where a physician uses complete history and examination results for diagnosis and treatment is seen as the next logical step. AI can do all these faster and probably with greater accuracy than humans due to their computing powers. A study was published in the Nature journal in 2022 which looked at mammograms using AI technology. The result revealed that there was a reduction in false negatives by 9.4 per cent and false positives by 5.7 per cent. This is very promising.
Disease diagnosis will probably benefit immensely from AI. There are various uses of AI in the diagnosis of cancer, cardiovascular and neurological diseases. Two things stand out in this regard. First, machine learning is already assisting in workflow management and decision-making, making the entire process efficient. Second, deep learning, another aspect of AI, can leverage various data mining techniques to identify embedded patterns. Both ML and DL will be critical for the diagnosis and prognosis of diseases. One field where changes are already prominent is Radiology. Computer-assisted diagnosis (CAD) is now quite common. IBM has developed a software called Avicenna, and it can scan diagnostic images and recognise abnormalities pretty well. Avicenna can also pull data from the available medical records, compare results and propose treatment options.
Diagnostic systems based on DL are already on the market. FDA approved Arterys, a medical imaging AI platform, in 2017, which can identify lesions, prepare reports, and present a list of potential diseases affecting the patient. There are AI systems being employed for colonoscopy and endoscopy to find out malignancies including pancreatic cancer. Data shows that DL-based technology evaluating chest X-rays can more accurately diagnose pneumonia than a human. An article in the Annals of Oncology journal (2018) presented a small dataset where AI diagnosed skin cancer more accurately than dermatologists.
The speed, efficiency and specificity of lab procedures can be tremendously improved by AI. This can significantly improve detection and quantification of microorganisms, disease classification and clinical outcome analysis. Things like blood culture and sensitivity testing have been impacted very positively with rapid results, allowing the selection of proper antibiotics. Timely administration of the right antibiotic is critical to achieve the eradication of microorganisms and reduce the risk of resistance.
AI can be very useful in increasing the efficiency of patient flow, thereby prioritising more important patients and reducing wait time overall. This is possible with AI-driven tools designed to interact with the patients, collect data and formulate a management plan. Such real-time inputs can significantly speed up decision-making of healthcare professionals.
Precision medicine should also benefit from AI. The whole concept of this approach is based on offering individualised treatment to each patient, according to their lifestyle, environment and genetic characteristics. As the speciality of AI is data mining and analysis at an extreme speed, you can imagine how efficiently we would be able to make the process. A group of researchers from the Georgia Institute of Technology conducted a study using patients' genetic data to train ML. The system was then able to predict the response to anti-cancer drugs with a high level of success.
Dixie B. Baker, a senior researcher in health information technology opined that with AI there will be a quicker turnout of large genetic, clinical, lifestyle and social datasets across a diverse population. Physicians will then just take a look at the screen and know what to prescribe. IBM's AI computer system Watson is a case in point. The Lineberger Comprehensive Cancer Center at the University of North Carolina fed Watson genomic data for more than 1000 patients. The objective was to test its ability to recommend treatment for them. Their verdict was Watson could find optimal treatments faster than a human.
Perhaps the most widespread but overlooked use of AI in healthcare is patient education. We already have AI-based chatbots providing diet charts, lifestyle advice and behavioural therapies. Some AI-powered apps can even help with lifestyle modification, e.g. smoking cessation. This is very important because patient adherence to treatment is largely dependent on how well they understand their disease and what is required of them.
Though AI will inevitably become a cornerstone of healthcare, there are challenges we must be ready for. AI-based systems rely on available data, and if the quality of the data is lacking then the outcome will be similarly poor. There are also questions about data privacy and security. To ensure security, robust cybersecurity measures are required. Continuous engagement between experts and the public is also crucial for safety. Incorporating AI into the existing healthcare system also means a huge financial investment, that not everyone is prepared for.
AI will transform healthcare. The process is already on the way, and we can only hope for a better future. Productivity and efficiency of the service delivery will be increased, and a better patient experience is expected. But that does not mean healthcare providers will be out of the picture. They will always be an integral part of the overall care to ensure patients are getting what they require.
imtiazdmc@gmail.com