Minimizing machine tool downtime is crucial in manufacturing due to its impact on production efficiency, profitability, and maintenance costs. Unplanned equipment failures result in production halts, high repair expenses, and prolonged downtime. This study investigates predictive maintenance strategies based on minimalist data (e.g., tool wear, temperature, and vibration) to forecast breakdowns, thereby reducing the reliance on complex and costly monitoring systems. Current AI and ML techniques often capture only correlations rather than true causal relationships, and they struggle with uncertainties in small datasets. To overcome these limitations, our approach employs a probabilistic framework that is rigorously tested to accurately reflect industrial conditions, enhancing both prediction accuracy and risk estimation. The ultimate goal is to develop a cost-effective and reliable solution for small and medium-sized enterprises, improving equipment management, reducing unplanned downtime, and optimizing maintenance planning.