Healthcare waste mismanagement persists because the organisational and communicative dynamics driving erroneous practices remain largely unaddressed, particularly in high-turnover clinical units that concentrate the highest volumes of hazardous waste and operational pressure. Existing training interventions typically address motivational, analytical, or operational dimensions in isolation, lacking a systemic and adaptive framework. This study designs and theoretically validates Maieutic-H, a multilevel training ecosystem integrating narrative and gamified motivational strategies, Video-Based Interaction Analysis of communicative practices, and an AI-driven support system employing an interpretable Random Forest classifier to prioritise corrective actions, embedded within a six-phase adaptive cycle with longitudinal monitoring at one, six, and twelve months. Theoretical validation through comparison with eleven programmes from the literature identifies three recurring, sub-optimal configurations—single-component, parallel-component, and quasi-integrated interventions—none of which combines data-driven prioritisation with interactional analysis to surface operational blind spots. Preliminary qualitative validation against expert interviews showed 93% concordance between operator-perceived priorities and model-generated relevance scores, informing an adaptive recalibration ( (Formula presented.) ) that weights field-derived evidence over the simulated training baseline. The Maieutic-H model offers a scalable, theoretically grounded framework for sustainable behavioural change and regulatory compliance in complex clinical environments, aligning interpretable machine learning with healthcare process engineering. This manuscript presents a model development and theoretical validation study. Evidence on effectiveness will require empirical testing of the full Maieutic-H pathway in hospital pilot studies.

The Maieutic-H Model: Integrating Machine Learning-Based Prioritisation, Interaction Analysis, and Motivational Strategies for Scalable Healthcare Waste Management

Marmora, Giovanni;De Feo, Giovanni
2026

Abstract

Healthcare waste mismanagement persists because the organisational and communicative dynamics driving erroneous practices remain largely unaddressed, particularly in high-turnover clinical units that concentrate the highest volumes of hazardous waste and operational pressure. Existing training interventions typically address motivational, analytical, or operational dimensions in isolation, lacking a systemic and adaptive framework. This study designs and theoretically validates Maieutic-H, a multilevel training ecosystem integrating narrative and gamified motivational strategies, Video-Based Interaction Analysis of communicative practices, and an AI-driven support system employing an interpretable Random Forest classifier to prioritise corrective actions, embedded within a six-phase adaptive cycle with longitudinal monitoring at one, six, and twelve months. Theoretical validation through comparison with eleven programmes from the literature identifies three recurring, sub-optimal configurations—single-component, parallel-component, and quasi-integrated interventions—none of which combines data-driven prioritisation with interactional analysis to surface operational blind spots. Preliminary qualitative validation against expert interviews showed 93% concordance between operator-perceived priorities and model-generated relevance scores, informing an adaptive recalibration ( (Formula presented.) ) that weights field-derived evidence over the simulated training baseline. The Maieutic-H model offers a scalable, theoretically grounded framework for sustainable behavioural change and regulatory compliance in complex clinical environments, aligning interpretable machine learning with healthcare process engineering. This manuscript presents a model development and theoretical validation study. Evidence on effectiveness will require empirical testing of the full Maieutic-H pathway in hospital pilot studies.
File in questo prodotto:
Non ci sono file associati a questo prodotto.

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11386/4960895
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 0
  • ???jsp.display-item.citation.isi??? ND
social impact