The Lost Promise of Digital Health
The digital transformation of healthcare has never quite delivered on its promise to revolutionise medicine. It ended up being more like some body else’s filing cabinet. You know all the right information is there but finding it can be very difficult. Electronic patient records were simply created to store and retrieve information. There was never any attempt to do prioritise, analyse or predict. Writing yearly reports remained the same chore of spending 90% of the time trying to find the data; in some ways no different to previous departmental paper records. However applying machine learning techniques may now finally allow the digital promise to be realised.
The Problem of Medical Clinics
Medical clinics are notorious for over running. The solution to this is often to reduce the number of patients that are seen, especially in specialties dealing with complex patients with perhaps multiple problems but this in turn drives up the time patients wait for specialist review. It is ironic however that much of this time is now spent looking through poorly coded notes in electronic patient records (EPR) rather than with the patient themselves. When reviewing a patient’s previous attendances at the hospital there is often no notation that details who saw the patient at that time or whether a specific medial letter contains vital information about a rare drug allergy, or a senior consultants serendipitous thoughts about a possible rare condition that has not been followed through. It is now common for letter not to contain all the previous medical history as this would simply be duplicating possibly hundreds of clinical letters. The job of the doctor seeing the patient is now as much about detective work to see what is contained in the medical records as it is about practicing medicine. The established mantra that every interaction in medicine needs to be adequately documented has also resulted in the unintended consequence of new electronic notes being retained for even the most trivial of interactions making it even easier to miss crucial information.
Medical big data
Medical information has now become a big data problem for which AI is eminently suited. Medical staff will now be presented with summary information and pertinent clinical data. Reinforced learning techniques will enable natural language-based note enquiry to interrogate complex notes. Medical letters from specific consultants will now be easily accessed, rare potential diagnoses now followed through and no need to repeat previously known medical histories as this will all be available. This will have a significant benefit to medical staff. The length of time they now spend looking through patient’s records will be dramatically reduced, and doctors can then spend their time with the patient. This has always been the crux of medicine; more time spent face to face leads to more patient engagement, more trust, more empathy and ultimately better-quality healthcare. It will also lead to better job satisfaction, a reduction in over working and therefore less burnout increasing staff retention – something that is absolutely vital when it is accepted that there are very few health care systems with adequate staffing.
The future
The use of AI to solve complex notes as a big data solution will only be the start. Once an individual note can be understood in this way, departments will be able to audit their practice and use AI to drive the quality care agenda. Patients with diagnosed conditions who have out of date investigations can now be automatically identified and the process of repeat investigation automated. The ingestion of expert working group clinical guidelines will also allow the automatic comparison of patient treatments to the accepted standards of care allowing out lying patients to be flagged up and consultations arranged. If the reason for the deviation from care is valid then this can be used as a reinforced learning tool. In pushing the barriers of medical knowledge, it will now be more possible to identify patients with specific traits to approach for medical trials and automate the process for inviting them to consider participation. The data that will be generated by such applications will ultimately allow AI to show its true face - the art of prediction. The key to future healthcare will be predicting development of disease with enough accuracy to allow preventative steps to be taken heralding the era of a entirely new chapter in medicine.
Neil Howell MB CHB PhD FRCS is an NHS consultant working in Cardiac Surgery. He first became interested in digital health and statistical modelling during his PhD and over the last year he has taken time to learn about artificial intelligence looking at the applications of reinforced machine learning and neural networks within the field of medicine and medical research. He believes that AI will change the face of medical practice as we know it and that whilst AI will not replace doctors, doctors using AI will replace those that do not.
By Neil Howell MB CHB PhD FRCS is an NHS Consultant working in Cardiac Surgery
**************************************************************************
As Neil identifies in his blog above, AI allows us to address old challenges in new ways. The effect of this is to move previously intractable problems to viable solutions. At ICS.AI we are working with Neil and others on building a range of practical AI solutions which form our Healthcare AI Transformation platform.
Neil will be talking about the Medical Notes Clinical Graph at our webinar on the 17th and 24th of September where ICS AI will be showing the Healthcare AI Transformation Platform featuring different Healthcare use cases. ” Martin Neale CEO, ICS AI Ltd.
For more information on our brand-new healthcare platform and how you can register to our upcoming webinars, click the link https://www.ics.ai/webinars