Scholarlink

<p>The increasing complexity of personalized healthcare necessitates intelligent recommendation systems capable of delivering accurate, patient-specific suggestions. However, traditional approaches such as content-based filtering and collaborative filtering often face challenges like data sparsity, cold-start problems, and limited contextual relevance. This study proposes a Hybrid Recommender System (HRS) that combines content-based and item-based collaborative filtering using a weighted fusion strategy to address these limitations. The system was evaluated using the MIMIC-III dataset with 10-fold cross-validation. Experimental results show that the proposed hybrid model achieves superior performance across all metrics, with a precision of 0.87, a recall of 0.91, and an F1 score of 0.89, outperforming Matrix Factorisation and Neighbourhood-based CF. Additionally, it achieves the lowest prediction errors, with MAE and RMSE reduced to 0.145 and 0.230, respectively, in 10-fold validation. These results confirm the model’s robustness, stability, and potential for real-world clinical deployment in personalized healthcare environments.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Hybrid recommender system for personalized healthcare: integration of content-based and collaborative filtering approaches

  • 期刊论文

  • 发表日期：

    2025年9月9日

查看全文
  • Shahnawaz Ahmad,
  • Mohd. Aquib Ansari,
  • Mohd. Arif,
  • Javed Ahmad,
  • Shabana Mehfuz,
  • Justin Onyarin Ogala,
  • O. S. Adewale

摘要

The increasing complexity of personalized healthcare necessitates intelligent recommendation systems capable of delivering accurate, patient-specific suggestions. However, traditional approaches such as content-based filtering and collaborative filtering often face challenges like data sparsity, cold-start problems, and limited contextual relevance. This study proposes a Hybrid Recommender System (HRS) that combines content-based and item-based collaborative filtering using a weighted fusion strategy to address these limitations. The system was evaluated using the MIMIC-III dataset with 10-fold cross-validation. Experimental results show that the proposed hybrid model achieves superior performance across all metrics, with a precision of 0.87, a recall of 0.91, and an F1 score of 0.89, outperforming Matrix Factorisation and Neighbourhood-based CF. Additionally, it achieves the lowest prediction errors, with MAE and RMSE reduced to 0.145 and 0.230, respectively, in 10-fold validation. These results confirm the model’s robustness, stability, and potential for real-world clinical deployment in personalized healthcare environments.