Bridging the Care Gap: An AI Dietitian for Sudan
In the bustling corridors of the Military Hospital in Khartoum, managing Type 2 Diabetes (T2DM) is a race against geography, culture, and a crushing shortage of specialists. An estimated 1 million people in Sudan live with diabetes, accounting for 10% of hospital admissions, yet rural provinces lack the dietitians needed to guide them.
For the 95% of Sudanese diabetics with Type 2, the primary medicine is the plate. However, standard Western diet plans fail to account for local staples like kissra, gorasa, and taglia.
The Intelligent Digital Specialist
To bridge this critical gap, researchers developed an "intelligent" medical expert system designed to act as a digital specialist. Built using Visual Prolog 5.2, it employs a forward-chaining inference engine.
This clinical tool translates a patient’s unique physical profile into a culturally tailored, five-meal daily plan. It mirrors the logic of a human specialist to provide immediate nutritional interventions where doctors are scarce.
Engineered Precision: The "GIGO" Resistance
The system's precision lies in its resistance to "Garbage In, Garbage Out" (GIGO). It uses a deterministic matrix, where a patient's Body Mass Index (BMI) and activity level calculate their Total Calories (TC).
Examples of the Caloric Multiplier:
- A Slim (BMI ≤18.5) patient with a Very Active lifestyle: Weight × 40
- An Obese (BMI ≥30) patient with Little Activity: Weight × 20
Built-In Safety Guards
The system acts as a proactive safety guard. If it detects signaled comorbidities like Gout or Heart Disease, it automatically triggers logic gates to restrict nutrients.
For instance, it can limit protein or milk intake to 2 servings, preventing dietary complications common in unsupervised patients.
A Culturally Relevant Database
The system’s utility is grounded in cultural relevance. Its database incorporates 45 specific Sudanese food items.
This ensures the "efficient treatment food" prescribed by the system’s logic aligns with the "interesting food" familiar and preferred by the patient.
The Path from Code to Clinic
Implementing this solution presents significant challenges, as noted by the researchers:
- Data acquisition in Sudan was "expensive" and hampered by a lack of available specialists during the 6-month development period.
- While the system automates metabolic math, longitudinal clinical trials to measure its impact on Hemoglobin A1c (HbA1c) levels have not yet been conducted.
- Full implementation of its Arabic-language interface and a rigorous maintenance phase remain for future work.
As global diabetic numbers are projected to hit 366 million by 2030, localized AI tools like this represent a vital shift. They move healthcare away from a "one size fits all" model and towards one that respects both the science of the body and the traditions of the table.
Based on study: "Development of an Expert System for Diabetic Type-2 Diet" by Ibrahim M. Ahmed and Abeer M. Mahmoud; International Journal of Computer Applications (December 2014).