Digital medicine and COVID-19: a reflection on the role of artificial intelligence
We are dominated by technology - phones beeping, the itch we feel in the morning to check our e-mail, the stress we feel when we are separated from our devices. Face-to-face contact and telephone conversations have been replaced by texting. Electronic medical records have replaced paper charts, and referencing medical apps is far more common than opening a textbook. There is no doubt that technology has revolutionized the way we live, and the way we practice medicine. We could pass days without interacting with others, and yet the world is more connected than ever before.
COVID-19 has heightened our dependence on technology, but in a way, this experience has made us nostalgic for exactly the opposite. It has turned our worship of technology upside down. Closed borders, city-wide lockdowns, forced social distancing, working from home; masks, face shields, double-gloving – the whole gamut of means we are using to protect ourselves has left me, and dare I say us, with a yearning to restore in-person human connections.
The psychosocial effects of quarantine are well documented: feelings of anxiety, detachment, isolation and loneliness1,2. With limited to no social contact, we resort to on-line communication platforms. With clinics being virtual and extra precautions taken at the hospital, we feel distant from our patients. The longer this goes on the more we miss the human touch; a hug from a friend, a handshake, the reassurance from holding somebody’s hand. This disconnect has made me think about the impact of technology in medicine, specifically as it relates to Artificial Intelligence (AI).
I do not profess to be an expert in AI – not even close. My knowledge is limited to casual reading and podcasts. We do know some advantages AI can provide: enhanced diagnostic specificity; increased efficiency; reduced physician burnout; and improved data sifting for rare diseases3,5. But can a robot “be human”? Does it have emotional intelligence? Some data suggests that through “sentiment analysis”, AI can detect underlying emotion and change its output accordingly4. This research is early in its course, and has many limitations including the heterogeneity of language, phrases or words with different meanings and the inherent “disorganized” nature of speech4.
The idea that AI could learn human emotion is exciting yet frightening. Yet the more I feel distant from my loved ones, the more I see us engaged in a forced social experiment testing the value of human connection. The COVID-19 pandemic experience has led me to hypothesize that our jobs will not become obsolete with AI. Our human nature is to want to be around other humans and nothing can replace this.
As physicians we are trained in the art and in the science of medicine – a phrase we are told on day 1 of medical school. Medical competency has expanded from its traditional role in technical expertise to include emotional intelligence. Specifically, we see the value of social context, empathy, and observation.
As sophisticated as they may be, a robot lacks social context5. AI depends on data collection and uses this data to continuously learn. People are more than technical data points – two people with the same disease, vital signs, age and gender are likely to differ in educational background, mental health, or socioeconomic status. It is hard, or rather it is unfortunately not routine, to collect data points on the social determinants of health necessary for AI to interpret data within a specific clinical context3. Humans have the advantage of seeing social context and using it to make medical decisions.
Countless times we use empathy to help patients through difficult times. We do this by sharing our own experiences, our own stories. Story sharing rather than story telling is uniquely human. It is hard to share with a computer. We can talk to a computer (think of all the recent Zoom meetings) but that virtual barrier remains a challenge to connecting with others on a deeper level.
Although COVID will allow us to re-think the advantages of virtual care medicine, some things require in person observation. Essential physical examinations require the eyes and hands of a health care provider. Examining a toddler for otitis media or auscultating through crying fits requires cooperation from parents, and creative distraction techniques with the toddler. Observation helps us pick up on body language, tailoring our discussion to meet the needs and reactions of our patients, but also to aid in diagnosis. AI can give us data points for this toddler - T 38.9, HR 180, RR 46, diaphoretic, maybe even pale – all signs that point to sepsis on paper; but humans synthesize this with our visual inputs – well perfused, well hydrated, crying, distractible – to counter the sepsis alert. Observation plus data is vital to our job and is not something (to which I am aware) yet capable of AI. It requires the human connection.
As we sit at home, experiencing loneliness together, we are living through a social experiment testing our need for human connection. COVID-19 has shaken our entire world and I hope we never forget this. I hope we learn from this, and that as providers we realize the impact we as human beings have on our patients. The digitalization of the world at this time has shown us that not everything can be artificial. Some things remain uniquely human.
Dr. Laura Betcherman is a third year paediatrics resident at the Hospital for Sick Children in Toronto. She will be starting her fellowship in nephrology.
References
1. Brooks SK, Webster RK, Smith LE, et al. The psychosocial impact of quarantine and how to reduce it: rapid review of the evidence. Lancet 2020;395:912-20.
2. Lai J, Simeng M, Wang Y, et al. Factors associated with mental health outcomes among health care workers exposed to Coronavirus Disease 2019. Jama Netw Open 2020;3(3):e203976.
3. Kerasidou A. Artificial intelligence and the ongoing need for empathy, compassion and trust in healthcare. Bull World Health Organ 2020; 98:245-50.
4. Chakriswaran P, Vincent DR, Srinivasan K et al. Emotion AI-driven sentiment analysis: a survey, future research directions, and open issues. Appl Sci 2019;9:5462.
5. Buch VH, Ahmed I and Maruthappu M. Artificial intelligence in medicine: current trends and future possibilities. Br J Gen Pract 2018;68(668):143-44.
