Artificial intelligence (AI) is entering the field of medicine and may soon help improve quality of care and the patient experience, one expert says.
AI is beginning to capture the knowledge of clinicians and combine it with best evidence, and patient specific data, says Guy Wood-Gush, MD, a neurosurgeon and ophthalmic surgeon, and CEO of Deontics, a provider of artificial intelligence for clinical pathways and clinical decision support systems.
“Prior technologies did not provide the capability to capture cognitive thinking that clinicians go through to manage complex clinical diagnosis and treatment, nor were patient preference taken into consideration,” Wood-Gush says. “New measurements and performance payment are being put in place that considers patient experience and satisfaction.”
AI also is responding to the need for transparency as patients want to know more about their care options, outcomes probabilities, and cost. This is especially true given the shifting insurance landscape and increasing health costs to individuals, he says.
“Patients want to explore what treatment options are available to them and if they are covered by their health plans. They want visibility to see where they are on their care journey and possible outcome scenarios in a simple, consumer-friendly way, avoiding medical jargon,” Wood-Gush says. “Using the right AI technologies and delivering them on multiple platforms can engage a patient throughout their care lifecycle. Patients need to be able to make decisions that may be against recommended procedures, based on their personal preferences. These need to be accounted for in the technology platform and presented in a clear and explicit fashion.”
Clinicians increasingly want technology to guide them through care treatment, especially in complex and overlapping disease cases, Wood-Gush says. These cases do not follow a single path and need to support nonlinear decision-making and argumentation, he says, and they need to be self-documenting such that results of a decision can be quantified. Then, the system can learn from decisions that are being made.
“A ‘thinking’ system that can augment clinician decision-making and provide outcome data, safety, quality, and cost, can become a learning health system over time as patterns are recognized and new best practices are established,” he explains. “Given the overwhelming and ever-changing amount of clinical evidence, it is becoming impossible for a human to keep up with it and apply to best decision-making. Clinicians need technology that will allow them to make better and more accurate diagnoses and search through relevant data.”
Ultimately, all constituents of the healthcare delivery process are focused on the same things — safety, quality, and cost control, Wood-Gush says. That calls for new, more advanced systems that can think like humans, manage mass amounts of data, and start to learn from experience, he says.
AI encompasses a broad set of different technical approaches and capabilities, which are complementary and capable of enabling these requirements. These include machine learning, knowledge representation, and reasoning, amongst others, and an ecosystem of small and large companies which represent all of these has now emerged capable of supplying very granular clinical decision support systems at the point of care, he explains.
“Machine learning and analytics technologies are becoming increasingly strong at recognizing patterns amongst data and highlighting diagnoses or probabilities of patients falling into different disease categories. Similarly, these technologies can evaluate outcomes data and feedback ‘learning’ to improve guidelines and treatment and patient management plans,” he says. “Reasoning technologies or AI can contextualize data at the point of decision-making at the point of care, identifying possible courses of action and showing arguments from the evidence base — typically national guidelines and local protocols, but also analytics outputs, genomics, and proteomics — to allow the physician and the patient to understand the pros and cons of different courses of action.”
AI will reason over the course of a patient’s disease in a dynamic manner, much like the GPS in a car, Wood-Gush explains. As patients enter data about their conditions, like blood sugar readings for a diabetic, the advice continuously updates, engaging the patient in decision-making about their care, and allowing them to fully understand the consequences of different possible treatment options.
The AI also would inform the patient on whether the care being offered is compliant with best evidence, and it will take part in shared decision-making with their physician on a much more even-handed basis than is otherwise possible.
“As the technology is not a decision tree technology, but much more sophisticated, it is able to get away from step-by-step algorithms, which physicians hate as they do not represent real life and cannot handle anything other than very simple decisions,” Wood-Gush says. “AI will improve the physician experience as well.”