Nowadays, clinical decisions are often correlated with uncertainty, mainly because of the characteristics of the data used. This data tends to be imperfect, which makes it challenging to manage (Owens and Sox, 2014), and it is typically diverse and produced by multiple loosely connected units, which makes it challenging to gather and systematize (Hammond, Jaffe, Cimino, & Huff, 2014, p. 212). According to Owens and Sox (2014), one of the approaches that can help professionals deal with the resulting uncertainty is probabilistic medical reasoning (p. 67).
Probabilistic medical reasoning involves a complex process of determining a pre-test probability of outcomes of a particular decision, gathering more data through a test, which can help to significantly reduce the uncertainty (or fail to do so), defining the post-test probability, and repeating the cycle if necessary. Thus, the primary function of biomedical data in this cycle consists of informing decisions. For instance, Baio and Dawid (2015) demonstrate the application of probabilistic analysis for health economic predictions, which, among other things, use biomedical data for their completion. Similarly, Tsoukalas, Albertson, and Tagkopoulos (2015) create a framework that uses biomedical data to assist in decision-making related to medication use and length-of-stay determination for patients with sepsis. To sum up, this application of biomedical data in probabilistic decision-making seems to be relatively common in various areas of healthcare.
Apart from that, biomedical data seems to have the function of informing the development of probabilistic models, and the two mentioned functions appear to be interconnected. For example, Mamiya et al. (2015) use biomedical information about newly diagnosed tuberculosis cases to develop a model, which can employ the biomedical information of future individual cases to determine the existence and length of the latent period of the disease. Similarly, Wang et al. (2014) use the existing information on acute myeloid leukemia to create a model that is dependent on various patient and treatment characteristics. That can predict life expectancy for patients with the illness, thus informing decision-making for future cases. To sum up, both functions of biomedical data in probabilistic decision-making are interconnected. They may be employed to develop and use probabilistic analysis tools to reduce uncertainty and provide a basis for decision-making in various healthcare settings.
Designing a Public Health Study
When considering the possibilities of a public health study, I would rather presume that I have significant resources that can be applied, for example, to the probabilistic analysis of spreading patterns of contagious diseases like tuberculosis. An example of such analysis is the work by Mamiya et al. (2015). The analysis is likely to be mixed methods or quantitative: qualitative data from interviews or questionnaires could be collected if necessary to complement quantitative from existing databases, electronic health records, and other sources. However, the analysis of the data would be carried out with the help of quantitative methods, that is, algorithms and models which are typically used in modern-day probabilistic studies. Such a study is only feasible if multiple sources of data are used, which makes it very resource- and time-consuming. Still, the results are likely to offer significant implications for the prevention and containment of contagious diseases, which is the most substantial and desirable outcome.
References
Baio, G., & Dawid, A. (2015). Probabilistic sensitivity analysis in health economics. Statistical Methods in Medical Research, 24(6), 615-634.
Hammond, W. E., Jaffe, C., Cimino, J. J., & Huff, S. (2014). Standards in biomedical informatics. In E. Shortliffe & J. Cimino (Eds.), Biomedical Informatics (pp. 211-254). London, UK: Springer.
Mamiya, H., Schwartzman, K., Verma, A., Jauvin, C., Behr, M., & Buckeridge, D. (2015). Towards probabilistic decision support in public health practice: Predicting recent transmission of tuberculosis from patient attributes. Journal of Biomedical Informatics, 53, 237-242.
Owens, D. K., & Sox, H. C. (2014). Biomedical decision making: Probabilistic clinical reasoning. In E. Shortliffe & J. Cimino (Eds.), Biomedical Informatics (pp. 67-108). London, UK: Springer.
Tsoukalas, A., Albertson, T., & Tagkopoulos, I. (2015). From data to optimal decision making: A data-driven, probabilistic machine learning approach to decision support for patients with sepsis. JMIR Medical Informatics, 3(1), 1-15.
Wang, H., Aas, E., Howell, D., Roman, E., Patmore, R., Jack, A., & Smith, A. (2014). Long-term medical costs and life expectancy of acute myeloid leukemia: A probabilistic decision model. Value in Health, 17(2), 205-214.