Over the past decades, science has advanced rapidly. We now know more about the human body than ever before, and modern medicine has advanced at a pace that allows healthcare professionals to administer treatments with a degree of precision that was unimaginable just a few years ago. But that knowledge has created a new degree of complexity because there is vastly more data potentially available from wearables, in addition to an increasingly detailed understanding of how the human body works. When the symptoms of several different illnesses present similarly and different treatment options can have different effects, how do doctors know the right combination of treatment options for a specific situation?
Today, it’s possible for patients to receive a highly personalised treatment plan based on their specific needs. One of the benefits of the vast recent expansion of medical knowledge is that treatments can be matched very precisely to patient needs.
Machine learning — the ability for a computer to identify a pattern and carry out a prescribed action — and artificial intelligence (AI) — where a computer can identify a pattern and decide what to do next — can help medical professionals determine the best clinical action to take with their patients.
For example, wearable devices and connected sensors can continuously collect data and send it to secure repositories. Until now, most measurements have been taken by a professional during an appointment, giving the doctor a point-in-time measurement but no longitudinal data. This makes it harder to identify trends and subtle changes in a patient’s condition.
But the availability of a huge amount of data poses a new problem: How can the data be analysed quickly and acted upon? Machine learning can be equipped with algorithms that can identify trends in data indicative of something requiring further action. For example, if a wearable device detects a subtle change in heart rate, this could create an alert that prompts a call to the patient for further investigation. This change might not be detectable if the patient is monitored only when s/he comes in for an appointment. By identifying a risk before it becomes acute, it becomes possible to intervene in a patient’s care before a severe incident occurs.
The other advance is the use of artificial intelligence as a diagnostic tool. For example, chatbots can be used to assist patients. Through an online chat with a bot, a person can work through a set of symptoms and then be directed to the best medical professional or service to help him or her. This doesn’t replace healthcare professionals. Rather, it supports them by assisting with the triage process and helps narrow down the next course of action that is needed.
Our knowledge of medical conditions is constantly expanding. For example, the traditional view of two types of diabetes is now being expanded to five different types. As a result, doctors need to be able to identify more issues for patients than before. This is another area where artificial intelligence and machine learning can assist.
When you pull together a set of test results, data from a wearable device or some other connected monitors, and a patient’s anecdotal information, the number of possible diagnoses may be significant. Multiplied by the potential number of different treatment options, ranging from behavioural changes to surgical intervention, the options available for a medical professional can be extensive.
Although AI and machine learning won’t replace the diagnostic process and decisions about treatment, these new technologies can help doctors make faster decisions by looking at all the diagnostic data from a patient, information about the latest drugs and treatment options, and the most recent research that is available.
Perhaps the fastest-growing area of medical research is in the field of genetics. This is where the intersection of medical science, data science, machine learning and AI are reaping significant rewards.
By using genomic information, it is possible to improve drug discovery processes and to better select patients’ cohorts for clinical trials. This means new treatments can be tested on the most appropriate patients. This complements current clinical trial mechanisms and, potentially, lets the approval process for candidates and researchers go forward in a faster and more efficient way.
In addition to genetic research, the study of how proteins are used in the body, proteomics, is evolving quickly. Scientists have known there is a close relationship between DNA and proteins and that the structure of a protein is tightly linked to its function. By using technology to better understand these structures and functions, new diagnostic tools, better medicines and enhanced treatment options can be developed that better match the individual needs of patients.
This is where AI and machine learning are leading the way. These, and other new technologies won’t replace medical professionals. But they will assist them with creating a more personalised treatment for patients. Instead of all patients of a similar age and with similar symptoms receiving the same treatment, we can use data to understand with greater specificity what is happening with that individual and offer a treatment plan that is highly customised to the person’s pathology, physiology and genetics.