Healthcare AI projects are characterized by high technological complexity, regulatory scrutiny, and socio-technical interdependencies, yet existing risk management frameworks offer limited guidance for addressing the distinctive uncertainties of adaptive, data-driven systems in multidisciplinary settings. This study examines how risks emerge and interact across the lifecycle of multidisciplinary AI healthcare projects and develops a lifecycle governance framework to support risk management in such environments. An exploratory qualitative case study was conducted within a multidisciplinary AI healthcare project in São Paulo, Brazil, involving a project team of 55 professionals. Data were collected through semi-structured interviews with 22 participants, document analysis, and direct observation. Data were analyzed through iterative thematic coding complemented by term frequency and co-occurrence analysis. Results indicate that technical, ethical, organizational, and regulatory risks are strongly interdependent and particularly concentrated in development and implementation phases, with data quality, algorithmic bias, privacy compliance, communication breakdowns, and resource constraints emerging as high-probability, high-impact risks. A governance gap was also identified: the absence of formalized risk processes and dedicated roles constrained traceability and strategic alignment despite the presence of informal mitigation practices. Based on these findings, a lifecycle governance framework is proposed to support the identification, prioritization, monitoring, and coordination of risks across AI project phases. Building on project risk management principles and AI governance literature, the proposed framework embeds lifecycle-oriented risk identification, prioritization, and monitoring within project governance structures. The study contributes to the literature by providing an empirically grounded framework for managing risk in multidisciplinary AI healthcare projects and offers practical guidance for organizations seeking to strengthen governance and decision-making throughout the AI lifecycle.