NEWS / Chatbot

Are Chatbots the Future of Healthcare?

July 16, 2019

As chatbots become more sophisticated and machine learning enables sideward steps between pre-set medical assessment algorithms, could chatbots save time and money by triaging patients before connecting them with the right specialist, wherever they are in the world?
…for medical professionals…
Many medical assessments have been refined into algorithms and calculators based on the best available evidence; for some conditions, doctors enter set information into a calculator to assess a patient’s likelihood of mortality after, for example, a cardiac event – the GRACE score – or calculate a patient’s likelihood of having a stroke, or whether the benefits outweigh the risks for a patient before prescribing, for example, blood-thinning medication. Medicine is a science based on research and knowledge, and medical education is almost analogous to uploading scenarios and flowcharts into the human brains of students, before new doctors are released into clinical practice to run these scenarios against real patients, effectively debugging themselves.
The calculators used to assess patient risk are based on huge global registries of events and outcome, and so are considered well-evidenced, good practice for use in practice. But do we really need doctors to spend their time filling in a form to create an automated score? With interoperable medical records including current observations and pathology results, this process could be automated. For assessment requiring a level of subjective human observation, this information could be entered into the records via a comprehensive chatbot, used either by the medical professionals or the patient themselves. A chatbot could instantly access medical records, blood results and more, run appropriate scenarios and ask questions. Even for trained human assessors, the use of mnemonics and set assessment patterns means that some medical assessments translate incredibly easy into chatbot algorithms.
…for everybody…
Many emergency health community health services now are accessed by phone call – often alongside the general number for emergency services is a number for people to call when they have non-emergency urgent care needs but are not sure where to go with them.
The people manning these phone lines do not necessarily have any medical background, but have set questionnaires to follow – specific questions to assess whether the patient needs emergency care, in which case they can trigger an ambulance response, and then a flowchart to follow to assess symptoms and make appropriate referrals. Any concerns can be escalated to a trained healthcare worker for a more in-depth assessment. The job of the first call handler; assessing urgency and need, making referrals or arranging help and possibly transport to hospital, is likely to be the first job to be delegated to a chatbot. Using a similar method, they could triage patients prior to general practitioner appointments, saving time for services that are already stretched. Some assessments will always need to be done face-to-face, human-to-human, but with rigorous pre-assessment and appropriate referrals, a person’s journey through the healthcare system could be smoother and quicker.
…and beyond.
With the future of interoperable systems monitoring IoMT devices in the home, calling a chatbot for medical advice will mean interacting with an intuitive interface that already has access – permissions will need to be ironed out – to all necessary medical history and results. When it comes to something as personal and critical as one’s health, there may be some natural reluctance to rely on a chatbot, to be able to open up about symptoms, or just to trust the system. Certainly in the early days of medical chatbot interfaces, it may be wise to give those seeking healthcare the choice to opt out and request a human operator. Although it may take some time, medical chatbots are likely to prove themselves an invaluable addition to frontline healthcare.