An international team of researchers examined how AI could help improve medication communications. In other words, they wanted to see how artificial intelligence could help patients and doctors have better conversations. The researchers wanted to find out, among other things, whether AI could detect high-risk situations when, for example, providers are under stress.
Medical practice may involve complex and stressful communication between patients and health care providers. Often, patients’ families also participate.
Healthcare providers often advise patients with chronic conditions to follow detailed treatment plans. Doctors, patients, and loved ones regularly have to make complex decisions during end-of-life care. Making these decisions can also be emotionally draining.
Treatment options for certain diseases are expanding rapidly.
How could treatment outcomes be improved if medical communications were more effective? Medical communications, in this context, refers to communication between healthcare providers, patients, and their families.
What is AI?
AI stands for Artificial Intelligence. It refers to software technologies that make devices think and behave like human beings. Devices include, for example, computers, medical devices, and robots.
AI contrasts with natural intelligence (what humans have). It is becoming more common in business, manufacturing, biomedicine, and many other sectors of the economy.
Some experts insist that it is only AI if it can perform at least as well as a human. In this context, ‘perform’ refers to our computational accuracy, capacity, and speed.
Humans can ‘learn on the job.’ In other words, we learn and adapt as we go along. Artificial intelligence also has this ability. We call it machine learning.
Researchers from the US, Ireland, and Scotland.
An international team of researchers explored the potential of AI to improve communications in healthcare settings. The researchers, from The Dartmouth Institute for Health Policy and Clinical Practice, Trinity College Dublin, and the University of Edinburgh, wrote about their study and findings in the BMJ (citation below). The authors were Padhraig Ryan, Saturnino Luz, Pierre Albert, Carl Vogel, Charles Normand, Edward Kennedy, and Glyn Elwyn.
Senior author, Glyn Elwyn, MD, PhD, MSc, a Professor at the Dartmouth, said:
“Many clinicians’ communications skills aren’t formerly assessed – either during school or in early practice. At the same time, there is a lot of evidence that clinicians often struggle when communicating with their patients.”
“It’s hard to improve on something when you’re not being given any feedback and don’t know how you’re doing.”
AI could potentially revolutionize medical communications
The authors say that AI has the potential to revolutionize medical communications. It could provide doctors with personalized, highly-detailed assessments of their communications skills. These assessments would cost considerably less than current methods which are employed sporadically.
The researchers point to three key areas of analysis in which artificial intelligence has the potential to improve medical communications significantly:
1. Analysis of words and phrases
AI could analyze words and phrases and offer feedback on whether providers and patients understood each other. The feedback could also determine how aligned they were in how they expressed themselves. It would tell us whether doctors had taken appropriate histories and offered evidence-based treatment.
The AI would also determine whether the provider spoke to the patient without using jargon.
In a press release, The Dartmouth Institute added:
“Eventually, AI also could be used to analyze conversations in real time and prompt providers to consider diagnoses that might not be obvious or to offer a wider range of treatment options.”
Turn-taking analysis
How much did the patient and provider speak, i.e., what proportion of the total speaking time? Did the provider pause enough so that the patient could ask questions or express concerns?
If patients get enough space to talk, they are more likely to take their medications according to the doctors’ instructions. They are also more likely to recall information
According to The Dartmouth Institute:
“The researchers say analysis of turn-taking could provide important insights into dialogue patterns and eventually intervene to prevent knee-jerk decisions to order invasive investigations.”
“For example, cases where more detailed questioning might have led to a diagnosis of heartburn rather than a presumption of cardiac pain.”
Tone and style in interactions
Algorithms are used by airlines to assess pilots’ key communication skills. The algorithms analyze their vocal energy and pitch.
By adapting such algorithms to medical communications, the AI might help detect high-risk situations. Doctors who are, for example, overworked or under a lot of stress, might communicate differently.
By analyzing patients’ voice patterns, the AI could provide information about their mental and physical health. Depressive episodes, for example, may be marked by systematic changes in vocal pitch. Some types of vocal changes could be an early indication of heart failure.
The Dartmouth Institute added the following comment regarding medical encounters and AI:
“The dialogue of medical encounters is complex. While a skilled provider will adjust their communication style to meet the needs of their patients, even the most advanced AI systems are incapable of parsing and assessing the complexities of these interactions – at least for now.”
Prof. Elwyn said:
“Five years ago, the idea of using AI to analyze medical communication wouldn’t have been on anyone’s radar. As the technology advances, it will be interesting to see whether healthcare systems can employ it effectively and whether providers will be open to using it as a tool for improving their communication skills.”
Citation
“Using artificial intelligence to assess clinicians’ communication skills,” Padhraig Ryan, Saturnino Luz, Pierre Albert, Carl Vogel, Charles Normand, Edward Kennedy, and Glyn Elwyn. BMJ 2019. DOI: https://doi.org/10.1136/bmj.l161.