Development and Field Testing of an Explainable Artificial Intelligence Clinical Decision Instrument for Stunting Prevention

Abstract

Artificial intelligence (AI) is increasingly integrated into clinical decision-making, yet its adoption in public health nutrition programs remains constrained by algorithmic opacity and limited healthcare professionals’ trust. This study aims to develop a human-centered Explainable Clinical Decision Intelligence framework to support trustworthy AI adoption in clinical decision-making for stunting prevention. The preparation and pilot-testing phase integrates Human–AI Trust Theory, Socio-Technical Systems Theory, Clinical Decision Intelligence, Information Processing Theory, Human-Centered AI, and Organizational Learning Theory. Twelve hypotheses were formulated for future covariance-based structural equation modeling (CB-SEM). A 35-item measurement instrument, purposive sampling strategy, informed consent protocol, and data collection procedures were developed and pilot-tested with 260 healthcare professionals from hospitals and primary healthcare centers. Descriptive statistics and preliminary reliability analyses revealed inadequate internal consistency across all seven constructs, indicating that the instrument required revision and re-validation before confirmatory factor analysis (CFA) and structural model testing. Consequently, no structural relationships were reported. Potential measurement deficiencies were examined, and a methodological roadmap for subsequent CFA and SEM analyses was established. The proposed framework integrates platform quality, explainable AI quality, decision transparency, healthcare professionals’ trust, decision acceptance, intervention quality, and stunting prevention effectiveness. Despite the current reliability limitations, this study establishes a theoretical and methodological foundation for future confirmatory research and offers guidance for ministries of health, healthcare institutions, clinicians, and health information system developers in implementing trustworthy explainable AI for clinical decision-making and stunting prevention.