Hybrid Artificial Intelligence Using Decision Tree and Heuristic Optimization for Healthcare Facility Recommendation Systems

Figure Ilustration AI

FORMOSA NEWS - Palembang - Hybrid AI System Revolutionizes Healthcare Facility Recommendations for Digital Health Applications. A team of computer science researchers from Universitas Multi Data Palembang has developed a breakthrough digital health solution that dramatically improves how patients find and select healthcare facilities. Published in May 2026, the study introduces an intelligent web-based recommendation platform that combines clinical triage logic with geographic optimization to ensure patients receive the right level of medical care at the right location. The pioneering research was conducted by a collaborative team at Universitas Multi Data Palembang, including Muhammad Junaidi, Abel Muhammad Zahrian, Satria Dimaz Mahendra, and Dicky Pratama. By deploying a Hybrid Artificial Intelligence framework, the researchers have successfully solved a major limitation in modern digital health systems: the inability of conventional mapping applications to account for medical urgency and facility capacity alongside simple geographic proximity.

The Digital Health Dilemma: Why Current Search Systems Fall Short
In an increasingly digitized world, finding a healthcare provider via smartphone or computer has become standard practice. However, conventional location-based search platforms generally provide static data sorted strictly by distance. They fail to evaluate individual user conditions, such as the severity of medical symptoms, physical mobility limitations, or the real-time operational density of nearby clinics and hospitals. This lack of intelligent data integration often leads to inefficient healthcare decision-making. For instance, a patient experiencing a severe medical emergency might unknowingly navigate to a small local clinic that lacks an intensive care unit. Conversely, individuals with minor ailments frequently crowd major hospital emergency rooms, causing unnecessary operational bottlenecks and administrative strain on critical care staff. To bridge this structural gap, the research team at Universitas Multi Data Palembang designed a unified intelligent framework. By merging classification and location optimization algorithms, their system ensures that healthcare recommendations are highly adaptive, personalized, and clinically relevant to each user's unique situation.

Deconstructing the Hybrid Artificial Intelligence Architecture
The core innovation developed at Universitas Multi Data Palembang relies on a dual-stage computational engine that processes user inputs through two distinct artificial intelligence methods:

• Medical Need Classification (Decision Tree): When a user accesses the web system, they input baseline parameters such as symptom severity, emergency level, age group, and mobility condition. The Decision Tree algorithm processes these attributes recursively to categorize the user’s exact medical need into one of three service tiers: Clinic, Hospital, or Emergency Unit.
  
• Geographic and Capacity Optimization (Heuristic Optimization): Once the correct tier of care is established, a heuristic optimization algorithm evaluates all matching facilities. This stage calculates near-optimal solutions based on a balance between travel distance and facility density.
  
• Mathematical Scoring Model: To guarantee precision, the system utilizes a specialized scoring formula that applies a 60% weight ($0.6$) to normalized distance and a 40% weight ($0.4$) to facility density. Because rapid accessibility is paramount during medical crises, distance is prioritized to keep travel times minimal while preventing patients from being sent to heavily overcrowded centers.
  

Experimental Results: Proving Efficiency Through Simulated Scenarios
To validate the efficiency of the hybrid model, Muhammad Junaidi and his co-authors utilized the structured Waterfall development model and evaluated their system against 150 simulated healthcare scenarios throughout 2026. The dataset mirrored real-world clinical variables and patient distribution challenges. The experimental results confirmed that the Decision Tree successfully generated transparent, interpretable rules that sorted patient needs accurately. Concurrently, the heuristic optimization engine successfully ranked competing facilities, assigning lower numerical scores to the most optimal choices. In one illustrative simulation comparing three distinct facilities, the system revealed the following performance outcomes:

• Facility A: Located at a distance of 2 kilometers with a 50% facility density, resulting in a final recommendation score of 1.40.
  
• Facility B: Located at a distance of 3 kilometers with a 20% facility density, resulting in a final recommendation score of 1.88.
  
• Facility C: Located at a distance of 1 kilometer with an 80% facility density, resulting in a final recommendation score of 0.92.
  

Because a lower score represents the most efficient compromise between short travel distance and adequate facility availability, the intelligent system automatically selected Facility C as the primary recommendation for the user.

Broad Implications for Smart Cities, Policymakers, and Society
The successful implementation of this hybrid artificial intelligence model marks a major leap forward for digital health informatics and smart city infrastructure. For individual citizens, the platform removes dangerous guesswork during unexpected medical events, offering clear directions to the most appropriate medical facility within seconds. For public health administrators and medical facility managers, the widespread adoption of such adaptive recommendation systems could optimize resource allocation across entire municipal networks. By intelligently diverting non-emergency patients away from overcrowded hospitals and toward underutilized community clinics, the system effectively balances patient loads, reduces wait times, and maximizes regional healthcare efficiency. Reflecting on the wider trajectory of this domain, lead researcher Muhammad Junaidi and his colleagues at Universitas Multi Data Palembang emphasized that combining multi-parameter optimization techniques within a single framework represents a vital evolutionary step over isolated, single-method healthcare applications. While this phase of the research was confined to a simulation-based dataset of 150 scenarios, it lays a robust foundation for scaling the system into real-world smart city environments using live geographic information and active patient registries.

Author Profiles
Muhammad Junaidi is a lead researcher affiliated with Universitas Multi Data Palembang. His primary area of expertise centers on digital health, medical informatics, and the practical application of Hybrid Artificial Intelligence systems.
Abel Muhammad Zahrian is a computer science researcher at Universitas Multi Data Palembang, specializing in supervised machine learning models, algorithmic design, and software engineering.
Satria Dimaz Mahendra is an expert in computational optimization frameworks, data analysis, and decision support systems at Universitas Multi Data Palembang.
Dicky Pratama is a research software engineer at Universitas Multi Data Palembang, focusing on web-based information systems and digital network architecture.

Source
Muhammad Junaidi, Abel Muhammad Zahrian, Satria Dimaz Mahendra, Dicky Pratama. Hybrid Artificial Intelligence Using Decision Tree and Heuristic Optimization for Healthcare Facility Recommendation Systems. Formosa Journal of Sustainable Research (FJSR). Vol. 5, No. 5 2026: 305-314.

DOI: https://doi.org/10.55927/fjsr.v5i5.32
URL: https://journalfjsr.my.id/index.php/fjsr