<p>Automatic text summarization is the process of identifying the most important information from a source document and compressing it into a shorter form while preserving the overall context. However, in biomedical text summarization, summarizing long documents is a challenging task due to the large volumes of information they contain. Additionally, the lack of supervised annotated data makes it difficult to design an effective summarization system. To overcome these challenges, we propose an efficient hybrid long-document summarization approach for biomedical documents. This approach enables the compression of large volumes of information through a two-stage process. In the extractive stage, we apply an ensemble technique to various proposed unsupervised summarization methods, which select the most representative sentences. These methods are based on diverse approaches, including UMLS-based, clustering-based, topic modeling-based, and graph-based extractive summarizers. In the abstractive phase, we propose a T5-based sentence compression technique to generate an abstractive summary. To evaluate the effectiveness of the proposed approach, we conduct extensive experiments on the PubMed biomedical summarization dataset. The experimental results show that our approach outperforms several baselines and achieves competitive performance to recent state-of-the-art methods in terms of ROUGE metrics.</p>