Small and Medium Enterprises (SMEs) in India face increasing cybersecurity threats due to limited resources, lack of awareness, inadequate infrastructure, and insufficient expertise. This study investigates the problems and challenges faced by SMEs in addressing cybersecurity incidents and evaluates the performance of various machine learning algorithms for predictive security analysis. A one- sample t-test was applied to analyze the significance of challenges, revealing that all identified factors—such as compliance issues, third-party risks, and insider threats—were statistically significant. To strengthen predictive security, multiple machine learning models, including REP Tree, J48, Multi-Class, AdaBoostM1, Logistic Regression, SMO, and Naïve Bayes, were compared against a Hybrid Meta Vote Predictive Model. The comparative analysis showed that the Hybrid Meta Vote Predictive Model outperformed others with the highest accuracy (67.87%), superior precision (0.85), and balanced recall (0.701), indicating its effectiveness in cybersecurity risk prediction. The findings emphasize the urgent need for SMEs to adopt advanced predictive models to address evolving cyber threats and enhance resilience, while also highlighting policy-level interventions to improve accessibility, awareness, and resource allocation for sustainable cybersecurity practices.