A Distributed Optimized Patient Scheduling using Partial Information

A software agent may be a member of a Multi-Agent System (MAS) which is collectively performing a range of complex and intelligent tasks. In the hospital, scheduling decisions are finding difficult to schedule because of the dynamic changes and distribution. In order to face this problem with dynamic changes in the hospital, a new method, Distributed Optimized Patient Scheduling with Grouping (DOPSG) has been proposed. The goal of this method is that there is no necessity for knowing patient agents information globally. With minimal information this method works effectively. Scheduling problem can be solved for multiple departments in the hospital. Patient agents have been scheduled to the resource agent based on the patient priority to reduce the waiting time of patient agent and to reduce idle time of resources.

💡 Research Summary

**
The paper addresses the challenging problem of patient scheduling in hospitals, where dynamic changes and distributed resources often lead to long waiting times and under‑utilised equipment. To cope with these issues, the authors propose a novel multi‑agent‑based framework called Distributed Optimized Patient Scheduling with Grouping (DOPSG). The key premise of DOPSG is that global knowledge of all patient agents is not required; instead, each resource agent (RA) operates with only locally available information and a limited set of data exchanged with its immediate neighbours.

The system architecture consists of three main components. The Master Health Check‑up Agent (MHCA) creates and registers patient agents (PA), maintains a count of active patients, and forwards patients to appropriate resource agents. Each resource (e.g., a diagnostic machine, a consultation room) hosts a Scheduling Agent (SA) that is responsible for triggering the scheduling process, selecting which patients to migrate, and executing the migration. Because every resource has its own SA, scheduling decisions are made in a fully distributed fashion without a central controller.

When a resource becomes overloaded (i.e., its current load exceeds its fixed capacity), it initiates a “migration set” protocol. The overloaded RA sends a request containing its identifier and the number of excess patients to a neighbouring RA. The neighbour checks whether its current capacity (Cj) is less than its fixed capacity (Fj). If there is enough space, the request is accepted; otherwise it is rejected and the overloaded RA contacts the next neighbour. Once a suitable target is found, the overloaded RA groups the excess patients into a set based on three criteria: (1) patient priority, (2) arrival time, and (3) the next required task. This grouping reduces the number of messages exchanged because the whole set is transferred as a single unit. The selected set is then migrated to the accepting RA, and the patients are removed from the source RA’s queue. The process repeats until all excess patients have been reassigned.

To evaluate the approach, the authors implemented the system using the Java Agent Development Framework (JADE) on a modest hardware platform (Intel Core 2 Duo T8100, 32‑bit OS). The experimental scenario involved three resources and fifty patient agents. Performance was measured using a suite of scheduling metrics commonly employed in manufacturing and healthcare: maximum completion time (Cmax), maximum tardiness (Tmax), total completion time (∑Cj), total tardiness (∑Tj), weighted completion time (∑WjCj), and weighted tardiness (∑WjTj). The results, presented in several graphs (e.g., Figure 5.1), indicate that DOPSG outperforms traditional First‑Come‑First‑Serve (FCFS) and Weighted Shortest Processing Time (WSPT) policies, achieving lower total completion times and reduced weighted tardiness. The authors attribute these gains to the reduced communication overhead (thanks to the migration‑set concept) and the priority‑aware grouping of patients.

Despite the promising outcomes, the study has notable limitations. The experimental scale (only three resources and fifty patients) is far smaller than a real‑world hospital, raising questions about scalability and robustness under heavy load. The comparison set is limited; state‑of‑the‑art approaches such as MedPaCo, MPAEX, and GPGP, which also use multi‑agent negotiation and auction mechanisms, are not included, making it difficult to assess the relative advantage of DOPSG. Moreover, the paper lacks a formal analysis of algorithmic complexity and does not quantify the exact amount of “partial information” required, leaving the trade‑off between information loss and scheduling quality unclear. Finally, the grouping criteria ignore many practical constraints (e.g., equipment maintenance, staff skill levels, patient‑specific clinical pathways), which could affect feasibility in a live clinical environment.

In summary, the paper introduces an interesting distributed scheduling paradigm that leverages minimal global data and a migration‑set strategy to alleviate resource congestion in hospitals. While the concept is sound and the initial results are encouraging, further work is needed to validate the approach at larger scales, to benchmark it against existing multi‑agent scheduling frameworks, and to incorporate richer clinical constraints. Such extensions would be essential for translating DOPSG from a simulation prototype into a deployable decision‑support tool for modern healthcare facilities.