A Transparent Rule-Based Decision-Support Prototype for Diabetic Menu Planning and Meal-Combination Optimization:...

Background/Objectives: Translating medical nutrition therapy guidance into practical meal plans requires consistent application of clinical and dietetic rules while accommodating different therapeutic regimens. This study developed and computationally evaluated a transparent Python-based decision-support prototype for diabetic menu planning. Methods: The prototype combines procedural decision rules, food-exchange patterns, nutritional checks, and mixed-integer linear programming (MILP). Two meal-frequency pathways were represented: a conventional multi-meal pathway and a three-main-meal pathway for people treated with modern basal-bolus insulin regimens. For the multi-meal pathway, a structured dataset containing 42 candidate meals (seven alternatives in each of six meal categories) was used for objective-specific optimization and exhaustive evaluation of all 117,649 possible daily combinations under predefined nutritional constraints. Sensitivity to the allowable energy deviation and trade-offs among nutritional objectives were also examined. Results: The prototype produced deterministic pathway selection, exchange allocation, nutritional summaries, and explicit warnings when required information was unavailable. Under the baseline ±6.5% energy tolerance, 5628 of 117,649 six-meal combinations (4.78%) satisfied all nutritional constraints. Objective-specific optimization identified feasible solutions for minimizing energy, carbohydrate, and saturated fat and maximizing dietary fiber. The proportion of feasible combinations ranged from 2.71% at ±5% energy tolerance to 9.47% at ±10%, while comparison of objective-specific solutions demonstrated measurable nutritional trade-offs. The three-meal basal-bolus pathway generated predefined exchange allocation and clinical-review warnings but was not subjected to the same exhaustive combinatorial optimization. Conclusions: The prototype demonstrates the computational feasibility of integrating explicit dietetic rules, regimen-related meal-pattern selection, food-exchange allocation, and mathematical optimization within an inspectable decision-support framework. The findings establish technical feasibility within the evaluated datasets rather than clinical effectiveness. Independent software testing, expert usability assessment, evaluation using larger and version-controlled meal datasets, and prospective clinical validation are required before routine clinical implementation.