Ludwig Wittgenstein wrote, “the limits of my language mean the limits of my world.” Fortunately, human communication carries multiple layers of signals – verbal and non-verbal, conscious and subconscious – with language or consciousness representing summarizations of all those signals. Similarly, the patient-physician interaction also is laden with signals. Even without considering the vagaries of memory, the ambiguities of language, the pressures of time and physician workflow, the distortions of implicit and explicit economic and legal pressures, the “subjective” aspects of medical documentation – the history of present illness, the review of systems, the physical exam, the medical decision making, diagnosis – are at its best, a coarse grained and one dimensional representation of this interaction. Important and useful information gets lost in these summations. Therefore, as electronic health record-based datasets generated during “normal” patient care becomes increasingly utilized to build tools that augment physician decision making, generate medical evidence, and improve physician performance, these datasets (even “big” ones) will lack important dimensionality, and at best lose important insight but at worst, systematize biases (future essay) and amplify unintended consequences.
A solution to minimize this limitation is to decompress and unpack medical documentation. Just like sensorization of the real world can complement patient intentions with actions by matching their interoceptive states with introspection and language (next essay), digitized audio and video feeds of the patient-physician interaction has the potential to disaggregate the conscious from the subconscious and the verbal from the nonverbal. The cadence, rhythm, and tone of speech, the give and take of conversational turn taking, the mimicry of facial expression, speech patterns, and posture all carry invaluable signals in human communication. Utilizing sensor technology, computational sociologist Alex Pentland, describes and validates four subconscious non-verbal signals that occur during a social interaction that have significant predictive and explanatory power. To contrast them with the intentionally planned conscious signals that are subject to manipulation and deception, he refers to these as honest signals: influence, mimicry, activity, and consistency.
Influence is defined as the extent by which one person can cause the other person’s pattern of speaking to match their own pattern. Influence serves to signal attention or how likely one is paying attention to a conversation. Mimicry as the reflexive copying of one person by another during a conversation, resulting in an unconscious back-and-forth trading of smiles, interactions, and head nodding during conversation is a proxy for empathy. Activity, which indicates interest and excitement, is tightly correlated with autonomic nervous system function. Consistency of emphasis in speech is a signal of mental focus, while greater variability may signal an openness to be influenced. For example, there is evidence that empathic physicians (mimicry) have improved outcomes. This becomes increasingly important as standardized treatment protocols are beginning to minimize variations in treatment between physicians. Top of mind and not by any stretch of imagination, I can also envision robust correlations between autonomic nervous system status (activity) with the quality of medical decision making, autonomic nervous system status and pain scores, and treatment compliance based on levels of influence.
In general, consciousness is a limited-capacity system and represents the proverbial tip of the iceberg of the mind. The vast majority of information from the world is processed below the level of consciousness. Not considering (for now) the ethical, legal, and economic concerns (future essay,) as technologies such as the sensorization movement continues and telemedicine, real-time dictation services, and Amazon’s Alexa gain traction in healthcare, documentation of the physician-patient interaction will be not only be automated, but the digitized and sensorized non-verbal signals and measured interoceptive states will be a rich source of data that will afford opportunities to build more refined predictive tools. In short, tapping into these signals will offer an opportunity to supplement the limits of consciousness and language with the expansive realm of the nonverbal and subconscious.
S-Fxn 