Artificial intelligence systems for complex decision-making in acute care medicine: a review
Tóm tắt
The integration of artificial intelligence (AI) into acute care brings a new source of intellectual thought to the bedside. This offers great potential for synergy between AI systems and the human intellect already delivering care. This much needed help should be embraced, if proven effective. However, there is a risk that the present role of physicians and nurses as the primary arbiters of acute care in hospitals may be overtaken by computers. While many argue that this transition is inevitable, the process of developing a formal plan to prevent the need to pass control of patient care to computers should not be further delayed. The first step in the interdiction process is to recognize; the limitations of existing hospital protocols, why we need AI in acute care, and finally how the focus of medical decision making will change with the integration of AI based analysis. The second step is to develop a strategy for changing the focus of medical education to empower physicians to maintain oversight of AI. Physicians, nurses, and experts in the field of safe hospital communication must control the transition to AI integrated care because there is significant risk during the transition period and much of this risk is subtle, unique to the hospital environment, and outside the expertise of AI designers. AI is needed in acute care because AI detects complex relational time-series patterns within datasets and this level of analysis transcends conventional threshold based analysis applied in hospital protocols in use today. For this reason medical education will have to change to provide healthcare workers with the ability to understand and over-read relational time pattern centered communications from AI. Medical education will need to place less emphasis on threshold decision making and a greater focus on detection, analysis, and the pathophysiologic basis of relational time patterns. This should be an early part of a medical student’s education because this is what their hospital companion (the AI) will be doing. Effective communication between human and artificial intelligence requires a common pattern centered knowledge base. Experts in safety focused human to human communication in hospitals should lead during this transition process.
Tài liệu tham khảo
Pauker SG, Kassirer JP. The threshold approach to clinical decision making. N Engl J Med. 1980;302(20):1109–17. https://doi.org/10.1056/NEJM198005153022003.
Vincent JL, Opal SM, Marshall JC, Tracey KJ. Sepsis definitions: time for change. Lancet. 2013;381(9868):774–5. https://doi.org/10.1016/S0140-6736(12)61815-7.
Lynn LA, Curry JP. Patterns of unexpected in-hospital deaths: a root causes analysis. Patient Saf Surg. 2011;5(1):3. https://doi.org/10.1186/1754-9493-5-3.
Lynn LA. Dynamic diagnostic relationism: a new diagnostic paradigm for complex rapidly changing clinical conditions. Patient Saf Surg. 2014;8(1):21.
Kent DM, Steyerberg E, van Klaveren D. Personalized evidence based medicine: predictive approaches to heterogeneous treatment effects. BMJ 2018;363:k4245. doi: https://doi.org/10.1136/bmj.k4245.
Nemati S, Holder A, Razmi F, Stanley MD, Clifford GD, Buchman TG. An Interpretable Machine Learning Model for Accurate Prediction of Sepsis in the ICU. Crit Care Med. 2018;46(4):547–53. https://doi.org/10.1097/CCM.0000000000002936.
Ching T, Himmelstein DS, Beaulieu-Jones BK, Kalinin AA, Do BT, Way GP, et al. Opportunities and obstacles for deep learning in biology and medicine. Journal of The Royal Society Interface. 2018;15(141):20170387.
Arora V, Johnson J, Lovinger D, Humphrey HJ, Meltzer DO. Communication failures in patient sign-out and suggestions for improvement: a critical incident analysis. Qual Saf Health Care. 2005;14(6):401–7. https://doi.org/10.1136/qshc.2005.015107.
Graham KL, Marcantonio ER, Huang GC, Yang J, Davis RB, Smith CC. Effect of a systems intervention on the quality and safety of patient handoffs in an internal medicine residency program. J Gen Intern Med. 2013;28(8):986–93. https://doi.org/10.1007/s11606-013-2391-7.