Progetto 1: Sequenziamento degli interventi chirurgici: un framework reattivo basato sulla programmazione lineare intera
Il progetto consiste nell'individuare la configurazione ottima dei Break-In-Moment (BIM), vale a dire dei tempi di rilascio delle sale operatorie che operano in parallelo durante l'arco della giornata. La minimizzazione dei Break-In-Interval (BII), vale a dire del tempo che intercorre tra due BIM consecutivi, permette infatti di garantire un rapido inserimento dei pazienti emergenziali. Verranno proposti modelli di Programmazione Lineare Intera per il sequenziamento delle sale operatorie prima dell'inizio dei blocchi operatori e durante (framework reattivo), che terrà conto dell'impiego delle risorse a disposizione (assegnamento paziente-equipe chirurgica e unità post-operatorie). Un'analisi di sensitività consentirà di studiare l'impatto del metodo proposto al variare dell'incertezza dei tempi operatori e del numero di pazienti emergenziali. Per questo progetto verranno usati open data e dati generati artificialmente, in modo tale da studiare l'effetto della metodologia proposta in diversi contesti operativi.
Progetto 2: Clusterizzazione outcome-aware dei percorsi clinici per la predizione della mortalità a breve termine
Il progetto mira a sviluppare un modello di clustering outcome-aware delle sequenze di cura ospedaliere dei pazienti con trauma maggiore, con l’obiettivo di individuare gruppi di percorsi clinici caratterizzati da differenti probabilità di mortalità a breve termine. Le sequenze di eventi (procedure chirurgiche, ricovero e trasferimenti nei reparti) vengono analizzate per estrarre pattern temporali e organizzativi rilevanti. Il modello integra tecniche di apprendimento supervisionato e clustering gerarchico per creare raggruppamenti che massimizzano la separazione in termini di esito clinico, fornendo così una base per comprendere quali configurazioni di percorso siano associate a esiti migliori o peggiori e supportare strategie di ottimizzazione dei processi assistenziali. Per il progetto verranno utilizzati i dati di un caso reale presso il Policlinico San Matteo di Pavia.
Project 1: Surgical case sequencing: a reactive framework based on integer linear programming
The project consists in identifying the optimal configuration of the Break-In-Moments (BIM), i.e. the release times of the operating rooms that operate in parallel throughout the day. The minimization of Break-In-Interval (BII), that is the time elapsing between two consecutive BIMs, allows the rapid insertion of emergency patients. Integer Linear Programming models will be proposed for the sequencing of operating rooms before the start of the operating blocks and during their execution (reactive framework), which will take into account the use of available resources (assignment of patient-surgical team and post-operative units). A sensitivity analysis will allow the study of the impact of the proposed method on varying the uncertainty of the operating times and the number of emergency patients. For this project, open data and artificially generated data will be used, in order to analyze the effect of the proposed methodology in different operational contexts.
Project 2: Outcome-aware clustering of clinical pathways for short-term mortality prediction
The project aims to develop an outcome-aware clustering model for analyzing hospital care sequences of patients with major trauma, with the goal of identifying groups of clinical pathways characterized by different probabilities of short-term mortality. Sequences of events (such as surgical procedures, hospital admissions, and ward transfers) are analyzed to extract relevant temporal and organizational patterns. The model integrates supervised learning and hierarchical clustering techniques to create groupings that maximize separation in terms of clinical outcomes, providing insights into which care pathway configurations are associated with better or worse outcomes and supporting strategies for optimizing clinical processes. Data from a real-world case at Policlinico San Matteo in Pavia will be used for the project.
Online algorithms for a general patient-centred radiotherapy scheduling problem (as a cosupervisor, by G. Squillace, M.Sc. in Computer Science, 2018)
A process discovery algorithm based on integer linear programming (by S. Biavaschi, B.Sc. in Mathematics, 2022)
Surgical case assignment and sequencing: a data-driven approach (by A. Daldossi, M.Sc. in Mathematics, 2024)
Patients who leave without being seen from the emergency department: a dynamic optimization approach based on survival regression (by V. Meini, M.Sc. in Mathematics, 2024)
Probabilistic algorithms for classifying data sets (as a cosupervisor, B.Sc. in Mathematics, by E. Faruffini, 2025)
Dual-horizon optimization of outpatient surgery scheduling: a Clustering and Stochastic Programming approach (by A. Salacrist, M.Sc. in Mathematics, 2025)
Optimizing intraday portfolio rebalancing in algorithmic trading (by M. Garbagnoli, M.Sc. in Finance, 2025)
Evaluation of diagnostic strategies in the Emergency Department via Discrete Event Simulation (by E. Dacrema, B.Sc. in Biomedical Engineering, 2025)
Matheuristic algorithms for the optimization of multivariate pairs trading (by T. Furas, M.Sc. in Finance, 2025)
The Class Constrained Multiple Knapsack Problem: Models, algorithms, and application to a real-world case study (by A. Cerbone, B.Sc. in Mathematics, 2025)
Machine learning-based confounding analysis of trauma mortality via optimal k-cardinality assignment (L. Terzi, M.Sc. in Mathematics, 2026)
Optimization of Computed Tomography examination scheduling through Integer Linear Programming with fairness criteria (by D. Demarchi, B.Sc. in Mathematics, 2026)
Extensions of the Paging Problem: Competitive Analysis and Experimental Evaluation in Medical 3D Printing (by F. Masnaghetti, M.Sc. in Mathematics, 2026)
Integer Linear Programming models for examination and reporting scheduling (by T. Bovati, B.Sc. in Bioengineering)
Learning-based online optimization of inpatient appointment scheduling (by A. Cima, M.Sc. in Mathematics)
Serious games for healthcare management (by J. Coroli, B.Sc. in Bioengineering)
Distributionally robust optimization for multi-resource operating room scheduling (by G. Costa, M.Sc. in Mathematics)
Genetic algorithms for parameter calibration in the synthetic generation of defective pharmaceutical product images (by C. Gallo, B.Sc. in Mathematics)
Optimization-guided learning for online operating room scheduling policies (by C. Marinelli, M.Sc. in Mathematics)
Optimization models for fair resource allocation in airport ground support operations (by T. Torrisi, M.Sc. in Mathematics)