Methods and formal models for healthcare systems management

Resumen

A healthcare system is an organization of people, institutions, and resources that deliver healthcare services to meet the health needs of target populations. The size of the systems, the huge number of agents involved and their different expectations make the management of healthcare systems a tough task which could be alleviated through the use of technology. In this thesis, new methods and formal models for healthcare system management are presented. Particularly, the thesis is divided in two main parts: the first one has to do with the modeling and analysis in hospitals by the use of clinical pathways while the second one deals with the planning and scheduling of patients in the operation rooms. Regarding the modeling and analysis of healthcare systems, depending on different visions and expectations, the system can be treated from different perspectives called facets. In chapter 2, the formal definition and characterization of two facets are given: (1) facet of resource management and (2) handshake between clinical pathways facet. They are obtained by applying to Stochastic Well-formed Nets (colored Petri Nets) modeling the healthcare system a set of relaxations, abstraction and modifications. In the first facet the subclass of S4PR is obtained which is a characteristic model of the resource allocation systems while in the second facet Deterministically Synchronized Sequential Process (DSSP) are considered. Both nets (S4PR and DSSP) are formal subclasses of Petri Nets where net level techniques can be applied. In chapters 3 and 4, we will focus on the liveness of the DSSP systems resulting from the facet of communication between clinical pathways. These kinds of nets are composed by agents (modeling clinical pathways) cooperating in a distributed way by the asynchronous messaging passing through the buffers (modeling the communication channels). In particular two approaches have been proposed. The idea behind the first approach is to advance the buffer consumption to the first conflict transition in the agents. Considering healthcare systems modeled by a DSSP, this means that before a patient starts a clinical pathway, all required information must be available. Unfortunately, this pre-assignment method only works in some particular DSSP structures which are characterized. A more general approach (than buffer pre-assignment) for liveness enforcing in non-live DSSP is given in Chapter. 4. The approach is formalized on two levels: execution and control. The execution level uses the original DSSP structure while for the control level we compute a new net system called the control PN. This net system is obtained from the original DSSP and has a predefined type of structure. The control PN will evolve synchronously with the non-live DSSP ensuring that the deadlock states will not be reached. The states (marking) of the control PN will enable or disable some transitions in the original DSSP, while some transitions in the control PN should fire synchronously with some transitions of the original DSSP. The second part of the thesis deals with surgery scheduling of patients in a hospital department. The Operating Rooms (ORs) are one of the most expensive material resources in hospitals, being the bottleneck of surgical services. Moreover, the aging population together with the improvement in surgical techniques are producing an increase in the demand for surgeries. So, the optimal use of the ORs time is crucial in healthcare service management. We focus on the planning and scheduling of patients in Spanish hospital departments considering its organizational structure particularities as well as the concerns and specifications of their doctors. In chapter 5, the scheduling of elective patients under ORs block booking is considered. The first criterion is to optimize the use of the OR, the second criterion is to prevent that the total available time in a block will be exceeded and the third criterion is to respect the preference order of the patient in the waiting list. Three different mathematical programming models for the scheduling of elective patients are proposed. These are combinatorial problems with high computational complexity, so three different heuristic solution methods are proposed and compared. The results show that a Mixed Integer Linear Programming (MILP) problem solved by Receding Horizon Strategy (RHS) obtains better scheduling in lowest time. Doctors using the MILP problem must fix an appropriate occupation rate for optimizing the use of the ORs but without exceeding the available time. This has two main problems: i) inexperienced doctors could find difficult to fix an appropriate occupation rate, and ii) the uncertain in the surgery durations (large standard deviation) could results in scheduling with an over/under utilization. In order to overcome these problems, a New Mixed-Integer Quadratic Constrained Programming (N-MIQCP) model is proposed. Considering some probabilistic concepts, quadratic constraints are included in N-MIQCP model to prevent the scheduling of blocks with a high risk of exceeding the available time. Two heuristic methods for solving the N-MIQCP problem are proposed and compared with other chance-constrained approaches in bibliography. The results conclude that the best schedulings are achieved using our Specific Heuristic Algorithm (SHA) due to similar occupation rates than using other approaches are obtained but our SHA respects much more the order of the patients in the waiting list. In chapter 6, a three steps approach is proposed for the combined scheduling of elective and urgent patients. In the first step, the elective patients are scheduled for a target Elective Surgery Time (EST) in the ORs, trying to respect the order of the patients on the waiting list. In the second one, the urgent patients are scheduled in the remaining time ensuring that an urgent patient does not wait more than 48 hours. Finally, in the third step, the surgeries assigned to each OR (elective and urgent) are sequenced in such a way that the maximum time that an emergency patient should wait is minimized. Considering realistic data, different policies of time reserved in the ORs for elective and urgent patients are evaluated. The results show that all ORs must be used to perform elective and urgent surgeries instead of reserving some ORs exclusively for one type of patient. Finally, in chapter 7 a software solution for surgery service management is given. A Decision Support System for elective surgery scheduling and a software tool called CIPLAN are proposed. The DSS use as core the SHA for the scheduling of elective patients, but it has other features related to the management of a surgery department. A software tool called CIPLAN which is based on the DSS is explained. The software tool has a friendly interface which has been developed in collaboration with doctors in the “Lozano Blesa” Hospital in Zaragoza. A real case study comparing the scheduling using the manual method with the scheduling obtained by using CIPLAN is discussed. The results show that 128.000 euros per year could be saved using CIPLAN in the mentioned hospital. Moreover, the use of the tool allows doctors to reduce the time spent in scheduling to use it medical tasks.