As artificial intelligence increasingly influences decision-making across healthcare, business, education, governance, finance, and public policy, the need for transparent, accountable, and trustworthy machine learning systems has become a critical scientific priority. While contemporary predictive models often achieve high predictive accuracy, their complexity can limit interpretability, hinder accountability, and reduce stakeholder trust, particularly in socially sensitive and cross-domain applications. This presentation introduces a new framework for trustworthy and interpretable AI that integrates statistical modeling, explainable machine learning, and accountability-oriented design to support responsible decision-making across diverse sectors. Central to the framework is icarm, a new R package developed by the author to operationalize interpretable and accountable machine learning workflows by combining data preprocessing, predictive modeling, fairness-oriented evaluation, explainability diagnostics, and transparent reporting within a unified analytical environment. Using models such as linear regression, random forests, support vector machines, neural networks, and gradient boosting, the package enables researchers and practitioners to develop high-performing yet interpretable predictive systems. This presentation demonstrates how icarm incorporates feature importance analysis, partial dependence profiling, local explanation methods, residual diagnostics, and model comparison tools to transform traditionally opaque machine learning models into transparent and actionable decision-support systems. Beyond technical interpretability, the framework emphasizes the role of explanations as instruments for accountability, policy communication, stakeholder engagement, and participatory understanding. By bridging statistical rigor, explainable AI, and human-centered design principles, this work contributes to the emerging field of trustworthy artificial intelligence and positions interpretability as a foundational requirement for ethical, scalable, and socially responsible AI deployment across civic and business domains.
