Prof. Jozo Dujmović
San Francisco State University, California, USA
Title: Graded Logic for Decision-Making in Digital Public Health
Abstract: Graded Logic (GL) is a continuum-valued propositional logic designed for modeling human commonsense reasoning and decision-making. GL is built upon graded truth values, graded simultaneity (graded conjunction), graded substitutability (graded disjunction), and the graded importance of logical variables. At its core is the Graded Conjunction/Disjunction (GCD) function, which unifies and balances conjunction and disjunction through adjustable parameters, allowing a seamless transition from strict simultaneity to complete substitutability. This flexibility forms the mathematical foundation of the Logic Scoring of Preference (LSP) decision method, widely used in decision-support systems [1, 2].
GL models of natural human reasoning are based on empirical observation, cognitive studies, expert evaluations, and behavioral data analysis, combined with mathematical formalization. GL principles are particularly relevant to professional decision support systems, computational intelligence, and artificial intelligence applications. One promising domain is digital public health, where GL and LSP can enhance decision-making in areas such as:
•Prioritization of limited medical resources: Vaccination scheduling in pandemics and organ transplantation priority evaluation.
•Medical condition assessment: Disease severity evaluation and patient disability measurement.
•Therapeutic decision-making: Optimizing risky therapies by balancing benefits and adverse effects.
•Healthcare system optimization: Selection of medical equipment and software, and strategic placement of hospitals, pharmacies, and medical facilities.
Let us briefly discuss each of these areas. One of the permanently present medical decision-making problems is the patient prioritizing problem that occurs whenever there are limited resources and many patients waiting for them. In such cases it is necessary to develop a justifiable criterion for computing the priority of each patient and then allocate resources according to decreasing priority degrees of patients (starting with the highest priority patient). Two typical such problems are the vaccination priority evaluation in pandemic conditions [3], and the organ transplantation priority evaluation [4, 2]. Of course, there are many similar problems of allocation of scarce resources, and the allocation of resources according to justifiable priority criteria yieldsthe optimum use of resources and maximizes positive public health effects. Generally, the criterion cannot be justifiable if it is incomplete. Thus, priority criteria should include both medical and social and/or ethical priority factors, as suggested in the liver transplantation prioritizing problem discussed in [4].
Disease severity evaluation is a typical evaluation decision problem [11]. This problem is significant because a quantitative indicator of disease severity is critical for deciding about the most appropriate therapy, and as the communication means between physicians (when a physician refers a patient to another physician). This is the reason why in many cases disease severity evaluation is based on rating scales [5-7] and on variety of scoring techniques [8-10]. In many cases it is possible to seriously criticize simple additive scoring used for disease severity evaluation and improve evaluation results using the LSP method [1].
Patient disability evaluation can significantly contribute to personalized medicine. Indeed, there are disabilities that are insignificant and negligible for a specific patient, but crucial for another patient. Reducing patient disability is the primary goal of all therapies, and in many cases that goal is only partially achieved. In many cases an overall disability degree can be computed using questionnaires posted on the Internet. Then both patients and physicians can monitor the status of disability as a function of time, assess the effects of a therapy, and make various other decisions. Such a system for monitoring the development of peripheral neuropathy can be found in [11, 1]. There are many specific diseases that can benefit from LSP models used for disability evaluation, and that is one of the promising areas for future research and development.
Risky therapy decision-making is an optimization problem. Some therapies are both beneficial and harmful. If a disease is increasingly developing, then it is desirable to start therapy as soon as possible. On the other hand, to reduce various negative effects of the therapy, it is desirable to postpone the therapy as late as possible. Obviously, between these extremes there is an optimum point that maximizes the patient’s quality of life. Such problems exist for many diseases and many therapies and deserve serious attention. A convincing solution of such a problem can be found in [1].
Evaluation and monitoring of the effects of physical (or another) therapy is necessary in almost all therapies. Physical (or other) therapies can prove their success by quantifying the attained decrease of patient disability. That is possible only if there is a quantitative disability criterion function that is updated during the therapy and used for explainable measurement of the degree of success. In the case of peripheral neuropathy, the use of such a system is described in [1]. Similar systems can be developed for many other therapies.
Selection of medical equipment and software tools is a problem like any other equipment or software selection problem. The LSP method is routinely used for evaluation, comparison and selection of complex equipment in electrical, computer, and software engineering [12-15]. The method determines the overall value of equipment/software as a logically justified combination of the overall suitability of equipment/software and its total cost. Evaluation, comparison, and selection of complex and expensive medical equipment can be based on the same LSP methodology.
Optimum location of hospitals, pharmacies, labs, and medical services is the same problem as the optimum location of other public objects (stores, theatres, sports stadiums, airports, etc.). Each point in a selected urban area has a degree of suitability as a location of a new medical institution. The LSP method provides location suitability maps, necessary for optimum decisions [1, 16-19].
This presentation is primarily based on presenter’s books [1, 2] and will cover the following topics:
(1) Properties of Graded Logic,
(2) GL-based professional decision-making method LSP,
(3) Applications of the LSP-based decision-support systems in digital public health:
(a) Prioritizing vaccination in pandemic conditions
(b) Prioritizing organ transplantation
(c) Evaluation of patient disability
(d) Solutions to the risky therapy decision problem.
Bio: To be announced soon.