Skip main navigation

Potential of AI in healthcare

This step introduces potential applications and outcomes of AI in various healthcare domains.
WHO guidance on Ethics & Governance of Artificial Intelligence for Health report front page
© AIProHealth Project

Now that you have already thought about the applications of artificial intelligence in healthcare yourself, let’s go over the main domains in which AI can be utilised. In this section we will rely on WHO guidance on Ethics & Governance of Artificial Intelligence for Health and summarise the major applications of AI in healthcare where it demonstrates promising potential.

Diagnosis and prediction-based diagnosis

Although not yet routinely used in clinical practice, AI systems often find their applications in supporting clinical diagnosis. With the progression of the technology, AI could help clinicians to make faster, more accurate diagnoses and avoid human error. In the situation of shortage of healthcare specialists, AI could be used to provide required assistance in patient assessment and diagnosing.

When a deep learning system is used to analyse images, you’re talking about computer vision. Computer vision is frequently used to analyse medical images such as MRIs, X-rays, or skin and eye imaging, to detect abnormalities. Because of the large amounts of medical data available in the field of radiology, AI is primarily being adapted and evaluated in this field, including radiological diagnosis in oncology. However, it’s also used in dermatology, pathology and ophthalmology. For example, computer vision can be used to identify and prevent conditions such as stroke, pneumonia, breast cancer and coronary heart disease.

AI may also be used to predict major health risks and help to prevent the onset of disease or mortality. For example, machine learning could be deployed to assess the relative risk of lifestyle diseases such as cardiovascular disease and diabetes.

Clinical care and clinical decision-making


Measuring blood pressure ©Storyblocks

AI can also be used to automate routine administrative tasks, such as filling in clinical reports, allowing doctors to spend more time with the patient.

One subfield of AI is called Natural Language Processing (NLP), which can also be used for speech and text recognition and processing. Audio signals, can be turned into words with the use of neural networks. The network can make predictions of the words that are being said at particular moments. In turn, a language model can go over these predictions and make corrections if necessary (for example, if two words are not likely to follow each other or if the grammar in a sentence is illogical).

Natural Language Processing can thus comprehend human speech and help to transcribe clinical notes for patient data entry. An example of when this can be useful is when a caregiver directly records a report after a visit to their client. This report can then automatically be turned into text. Another example is the transcription of the conversation between a doctor and their patient during the consultation. Moreover, this automatic transcription can be very useful in situations in which the healthcare professional is not able to write down observations, such as in the operating room or in a sterile lab.

In turn, these large amounts of unstructured patient data (texts) can be analyzed and categorized using NLP information extraction techniques. This can be used to fill in the data at the right place in the patient’s dossier.

Using the collected data of the patient, clinicians might use AI to extract information regarding procedures, therapies and the corresponding codes. AI can also be used to integrate and summarise various clinical notes, identify patients with specific risks. By using AI, clinicians receive support in difficult treatment decisions and catching clinical errors. AI-based clinical decision support systems (CDSS) as an advisory tool can also use this data to assist medical professionals in making better-informed decisions, e.g. recommend medications and doses, and suggest periodic follow-up checks and tests to ensure optimal patient care.

Another way to use AI in clinical care is with the use of robots. Robots that combine computer vision with reinforcement learning can support surgeons in the most dangerous aspects of the surgery they need to perform. Furthermore, AI can also be used for social robots, which use natural language processing to be able to communicate with a patient. Additionally, they can be used to support patients in their daily tasks such as bringing someone to the toilet or administering medicine.

Health systems management and planning


Traditional planning to be replaced by AI ©Storyblocks

Potential utilization of AI in health systems management include scheduling patients, estimating the length of stay in the hospital, predicting patients that are unlikely to attend a scheduled appointment, and allocation of health-system resources according to current challenges.

AI can be used to assist personnel in complex logistical tasks, such as optimization of the medical supply chain, performing tedious routine tasks or supporting complex decision-making. Adaptation of AI CDSS can help to cut costs of the medical provider by optimising the course of treatment, and the number of procedures, suggesting best suitable medications and reducing the time spent by clinicians on administrative paperwork.

In public health and public health surveillance

AI-based techniques can improve public health surveillance (including prediction-based surveillance) including identification of disease outbreak and response. For example, NLP algorithms can be used to analyse real world data gathered from news outlets, public health resources and social media for the mentioning of high-priority diseases. Additionally, tools can be used to analyse air-travel data to assess the risk of people arriving and departing being infected with those diseases. Furthermore, when looking for the onset of new diseases, unsupervised learning could be used on pooled data from different types of care facilities (e.g. hospitals, family doctors, etc.) to identify patterns in data that could indicate an upcoming epidemic.

AI in research, personalised medicine and drug development

AI can leverage large amounts of complex patient health data together with genetic information to help guide clinical care and personalized approaches to diagnosis and treatment. Personalized medicine is individually tailored to a person’s genes, lifestyle and environment. It can include development of disease risk-scores prediction models and pharmacogenetic applications that aim to provide the best suitable medication based on the person’s genetic information on how (s)he will respond to it.

In the time and cost consuming process of drug discovery and development, AI could be used to shorten this process and make it less expensive and more effective. AI can assist in structure-based drug discovery by predicting the 3D protein structure and model the interaction with the target protein, thus predicting the effect of a drug compound on the target before drug synthesis and production. Machine learning models have also been used for drug repurposing and discovering drug efficacy.

Both of these fields are intensive areas of research with lots of efforts made to bring developed AI solutions to real life applications.

Patient self-management of health


Self-management ©Storyblocks

Finally, AI tools may support the management of patients’ health outside clinical settings. It is especially important for patients with chronic diseases such as cardiovascular diseases, diabetes or mental health problems. AI could support self-management by using conversation agents, known as “chatbots”, health monitoring and risk prediction tools. Wearable devices including those placed in the body, such as artificial limbs and smart implants, on the body, such as insulin pump patches, or near the body, such as activity trackers and smart watches capture large amounts of data. This creates more opportunities to recruit AI for monitoring a person’s health and for predicting health risks in a more prompt and efficient manner and taking actions such as notifying a health-care provider of any emerging concern.

Conclusion

As you could observe, there are quite some possible applications of AI in different aspects of healthcare. AI is not meant to substitute clinicians but it could add value to the everyday practice of healthcare professionals and patients by enhancing diagnostics, by helping to support decisions made by clinicians and by improving patient care. You will learn about examples of AI in healthcare in the next video.

© AIProHealth Project
This article is from the free online

How Artificial Intelligence Can Support Healthcare

Created by
FutureLearn - Learning For Life

Our purpose is to transform access to education.

We offer a diverse selection of courses from leading universities and cultural institutions from around the world. These are delivered one step at a time, and are accessible on mobile, tablet and desktop, so you can fit learning around your life.

We believe learning should be an enjoyable, social experience, so our courses offer the opportunity to discuss what you’re learning with others as you go, helping you make fresh discoveries and form new ideas.
You can unlock new opportunities with unlimited access to hundreds of online short courses for a year by subscribing to our Unlimited package. Build your knowledge with top universities and organisations.

Learn more about how FutureLearn is transforming access to education