The ways AI is transforming healthcare

Artificial Intelligence (AI) is rapidly becoming a hot topic in the healthcare industry. The healthcare world’s excitement in AI is driven by the belief that AI has enormous potential to improve human health while simultaneously reducing healthcare costs. This means that advanced healthcare could soon become more accessible to everyone on the planet. AI has already shown promise in areas such as disease diagnosis and drug discovery, and additional health-related applications of AI are constantly being explored. To illustrate just how substantial AI’s presence in healthcare may soon be, consider the following two estimates: 1)    Market research firm Global Market Insights believes healthcare-related AI businesses could be worth $10 billion globally by the year 2024. 2)    Consulting firm Accenture estimates that AI-related healthcare cost savings could reach $150 billion annually in the US by 2026. Given such estimates, it’s no wonder AI is one of the hottest areas in healthcare. In this article, I explore the current areas AI is making the biggest impact on healthcare.

Applications of AI in healthcare

Disease Diagnosis

Inaccurate and delayed diagnoses can harm human health and increase healthcare costs. As such, there is enormous interest in improving diagnostic accuracy and speed. Companies have already begun to explore AI as a tool for improving diagnoses. UK company Babylon Health is currently testing a mobile app to help provide initial screening diagnoses. When patients come down with symptoms, they can enter their symptoms in to Babylon’s app. The app then compares the patient’s symptoms against symptom patterns, derived from deep learning methods, that are indicative of various medical conditions. Based on this comparison, the app diagnoses the patient’s condition and recommends the next step. The hope is that Babylon’s app, by being able to assess symptom patterns more thoroughly and quickly than a human doctor, will lead to more accurate and timely diagnoses.

Medical Image Processing and Interpretation

Medical imaging is a key part of disease diagnosis. Although medical images can be useful diagnostic aids, they can be prone to human error in a number of areas, including how the images are recorded, processed, and interpreted. Companies have begun to use deep learning as a method for making the medical imaging process more reliable and less prone to human error. For example, UK company Intelligent Ultrasound (recently acquired by Metaphor), has shown that images of fetal thighbones acquired and interpreted with the assistance of deep learning are significantly more accurate than images obtained by conventional ultrasound methods. Such results suggest deep learning has great promise as a method to improve the accuracy of medical imaging.

Precision/Personalised Medical Treatment

Once a patient has been diagnosed with a condition, the clinician must decide upon a treatment plan. Often, there are multiple treatment options, but it is not clear to the clinician which treatment option is best for the particular patient. IBM’s Watson Health is attempting to use AI to assist clinicians with treatment decisions in the area of oncology. Their system uses deep learning to discover patterns which indicate how various types of patients respond to various oncology treatments. Based on these patterns, the system can suggest to the clinician the treatment option which has the highest likelihood of benefitting the clinician’s specific patient.

Improvements in Administrative Processing

Managing patient care involves a great deal of record keeping and administration, which can be prone to error and is time-consuming. Nuance Communications is a company that believes AI can help streamline the burden of medical administration and record keeping. Nuance has developed a virtual clinical assistant which uses speech recognition and natural language processing to record and understand a clinician’s verbal encounter with a patient. The recording is translated into digital form and combined with other information in the patient’s electronic records. Additionally, their virtual assistant uses machine learning to detect common clinical patterns within the electronic medical record and prompts clinicians to record data necessary to document that clinical pattern. Impressively, Nuance reports that their virtual assistant improves the quality of clinical documentation by 36% and improves the speed at which clinical records are recorded by 45%.

Drug Discovery

Scientists are constantly searching for new and better drug treatments. But in order to identify opportunities for new drug treatments, scientists must have a good understanding of existing treatments and research. Unfortunately, keeping abreast of existing treatments and research is more and more difficult given the rapid rate at which scientific and medical studies are being published. Indeed, it is estimated that a new biomedical study is published every 30 seconds. To solve this problem, UK company BenevolentAI (formerly Stratified Medical) has begun to scan the biomedical literature using AI and machine learning. This allows the literature to be scanned more quickly and comprehensively than can be done by humans. It also facilitates the identification of patterns among studies that could provide novel insights about which chemical compounds might be promising drug candidates. Benevolent AI has begun to use these techniques to search the literature for promising compounds to treat ALS, and has already had some initial success in this area.

AI-Assisted Surgery

Recently, there has been interest in using deep learning in surgery. In particular, there is a belief that deep learning could help provide real time information that could inform surgical procedures and make them easier to conduct. As an example, deep learning image processing could help a surgeon detect, in real time, when he or she has encountered the boundary of a tumor. Although the use of deep learning in surgery is still in the early stages, companies such as Verb Surgical are actively working in this area.

The Future of AI in Healthcare

As computing power increases, it will become easier to implement computationally intensive machine learning and deep learning techniques. Therefore, assuming these techniques are proven to improve health outcomes and lower costs, they will likely become increasingly common in the healthcare industry. Indeed, rather than being the exception, AI-driven approaches may eventually become the standard by which many areas of the healthcare system operate.     If you would like to know more about AI in healthcare we have written an article specifically focused on how computer vision and natural language processing are revolutionising healthcare. You can read this here.

If you have enjoyed this article, please consider signing up to receive our monthly newsletter below which includes our latest articles, jobs and industry news.

View our latest selection of jobs