The constant search for resource optimization without impacting the quality of delivery of health services has been the premise of advances in the management literature applied to health and, also, of advanced models adopted in health systems in multiple countries 83,84. To achieve excellence, the management of health systems and services requires that managers have the opportunity to use and analyze data of good quality.85 The ability to generate and use accurate health data contributes to assertive and agile decision-making in the systemic management of health services and technologies 86. To formulate health policies and adopt good strategies, managers need to identify, analyze and solve problems and implement effective and timely actions. There is a demand for efficient allocation of resources and improvement in the quality of services, promoting greater organizational performance 87 and deliverables with added value. Therefore, it is imperative to produce useful information and knowledge to guide the action and support the different levels of decision, in order to allow the use of the information and knowledge produced in the activities of recognition, planning and definition of priorities, as it expands access to information and knowledge through the use of secondary data sources.
The health management field has been modernizing and the AI can optimize processes, reduce costs and improve planning. These resources facilitate the selection and training of managers, defining an appropriate set of skills and competences that, in turn, can contribute to efficiency, effectiveness and responsiveness in health organizations. Countless reports of management support systems are delivered using AI 86,88. Huang and Wu, in 2017,89 developed a model capable of predicting the number of outpatient appointments in an urban hospital and helping managers to manage hospitals more effectively. Using the database, it is possible to identify cases of greater risk for hospitalization or seek emergency care and to define more individualized schedules for reconsultation. Through natural language processing of medical notes, several algorithms have been used to identify complications and outcomes with excellent accuracy. In Brazil, a study used data from healthcare facilities and developed a model to compare program options designed to increase the efficiency of the healthcare workforce.90 Recently, AI techniques have been employed to estimate bed demands for COVID, equipment demands for use in ICUs, respirators and health professionals.91
AI employment may guide managers’ choices in allocating resources, defining priorities, identifying vulnerable populations to guide intervention and investment in health. One can mention, for example, the creation of the Health Vulnerability Index (IVS), in Belo Horizonte, to subsidize the planning of health actions, the allocation of resources, the implementation of the Family Health Strategy, and human resources in health. The IVS, created with census information for each census sector, includes the following components: (1) sanitation conditions (sewage, water supply and waste disposal), (2) housing, (3) education, (4) income , and is aggregated into 4 risk strata: low, medium, high and very high (Braga, 2010). These indicators and other indexes can be automated for all Brazilian municipalities and support management, setting health priorities, allocating resources, among others. The development of predictive models that allow generating information about the highest and lowest chances of hospitalization of patients helps hospitals and operators to better measure capacity and remuneration. However, the success of these solutions requires databases to be integrated 92,93 so that greater traceability of states and patients’ health history would be possible. The access to data on comorbidities and patient outcomes can contribute to the system being better sized from an economic and financial perspective.94 For remuneration policies to approximate models that pay for added value, it is necessary to be able to accurately monitor the way in which health services are delivered, considering the individual clinical conditions of each patient 95. In this sense, natural language processing algorithms have been used as a solution for the automation of classification of textual data 96 of health records of patients 97.
All of these solutions, when in operation, also allow the flow management of patients to be more easily dimensioned. Solutions that help the right patient to arrive at the right time in the right place contribute to the initial premise of health service management to deliver the best health results without exponentially increasing costs. Brazil invests approximately 8% of the GNP in the health system, 4% of which in the national Unified Health System (SUS – Sistema Unificado de Saúde), which covers 75% of the population. The adequate quantification and distribution of these resources needs to be as efficient as possible so that the majority of the population is able to access the health system in an appropriate manner, and in conditions to achieve better outcomes. For this reason, an agile flow management has a great responsibility in helping patients to be prioritized considering their specificities. Such flow management can be built based on AI models when databases start to be integrated and monitor comorbidities and outcomes 98.
Finally, given the sensitivity and impacts that managerial decisions have on the respective Health Systems, it is essential that issues related to ethics, transparency, privacy, responsibility, explainability and reliability are considered when talking about decision making based on data and AI algorithms 88. Managers need to trust the data and suggestions/analyzes of the system to feel comfortable in subsidizing his/her decisions in automatic suggestions. It is necessary to understand the logic behind the algorithmic suggestion/analysis, which needs to be self-explanatory, because in the last case, the manager can/should be charged in his decisions when accepting (or not) these algorithmic suggestions/analyzes, under the point from an ethical, legal, accountability, transparency and, ultimately, auditability and governance.
Principal Researchers: Carisi Polanczyk, Antonio Ribeiro, Altigran Silva, Marcos Gonçalves
References:
83. Porter ME. What is value in health care? N Engl J Med. 2010 Dec 23;363(26):2477–81.
84. Lakdawalla DN, Doshi JA, Garrison LP Jr, Phelps CE, Basu A, Danzon PM. Defining Elements of Value in Health Care-A Health Economics Approach: An ISPOR Special Task Force Report [3]. Value Health. 2018 Feb;21(2):131–9.
85. Dafny LS, Lee TH. Health care needs real competition. 2016; Available from: https://www.hbs.edu/faculty/Pages/item.aspx?num=51981
86. Gopal G, Suter-Crazzolara C, Toldo L, Eberhardt W. Digital transformation in healthcare – architectures of present and future information technologies. Clin Chem Lab Med. 2019 Feb 25;57(3):328–35.
87. Stefl ME. Common competencies for all healthcare managers: the Healthcare Leadership Alliance model. J Healthc Manag. 2008 Nov;53(6):360–73; discussion 374.
88. Perrault R, Shoham Y, Brynjolfsson E, Clark J, Etchemendy J, Grosz B, et al. The AI Index 2019 Annual Report. AI Index Steering Committee, Human-Centered AI Institute, Stanford University, Stanford, CA. 2019;
89. Huang D, Wu Z. Forecasting outpatient visits using empirical mode decomposition coupled with back-propagation artificial neural networks optimized by particle swarm optimization. PLoS One. 2017 Feb 21;12(2):e0172539.
90. Yousefi M, Yousefi M, Ferreira RPM, Kim JH, Fogliatto FS. Chaotic genetic algorithm and Adaboost ensemble metamodeling approach for optimum resource planning in emergency departments. Artif Intell Med. 2018 Jan;84:23–33.
91. Castro MC, de Carvalho LR, Chin T, Kahn R, Franca GVA, Macario EM, et al. Demand for hospitalization services for COVID-19 patients in Brazil. MedRxiv. 2020;
92. Halevy A, Norvig P, Pereira F. The Unreasonable Effectiveness of Data. IEEE Intell Syst. 2009 Mar;24(2):8–12.
93. dal Bianco G, Gonçalves MA, Duarte D. BLOSS: Effective meta-blocking with almost no effort. Inf Syst. 2018 Jun 1;75:75–89.
94. Feeley TW, Mohta NS. Transitioning payment models: fee-for-service to value-based care. 2018; Available from: https://www.hbs.edu/faculty/Pages/item.aspx?num=55281
95. van Egdom LSE, Lagendijk M, van der Kemp MH, van Dam JH, Mureau MAM, Hazelzet JA, et al. Implementation of Value Based Breast Cancer Care. Eur J Surg Oncol. 2019 Jul;45(7):1163–70.
96. Cunha W, Canuto S, Viegas F, Salles T, Gomes C, Mangaravite V, et al. Extended pre-processing pipeline for text classification: On the role of meta-feature representations, sparsification and selective sampling [Internet]. Vol. 57, Information Processing & Management. 2020. p. 102263. Available from: http://dx.doi.org/10.1016/j.ipm.2020.102263
97. Velupillai S, Suominen H, Liakata M, Roberts A, Shah AD, Morley K, et al. Using clinical Natural Language Processing for health outcomes research: Overview and actionable suggestions for future advances. J Biomed Inform. 2018 Dec;88:11–9.
98. Aboelmaged M, Hashem G. RFID application in patient and medical asset operations management: A technology, organizational and environmental (TOE) perspective into key enablers and impediments. Int J Med Inform. 2018 Oct;118:58–64.
