Scholarlink

<p>Recognizing the paramount importance of accurate fecundity prediction in assessing pregnancy possibilities, this research aimed to contribute significantly to healthcare methodologies by introducing a novel methodology in fecundity time series data analysis. While previous endeavors incorporated deep learning, notably LSTM models, integrating subfertility concepts, their limitations in handling extensive global data dependencies posed challenges to prediction accuracy. This study proposed a unique approach—fusing LSTM, termed as LSTMUE in the fecundity analysis domain, into a transformer encoder architecture with unique attention head selection mechanism. The evaluation of this proposed methodology utilized two distinct fecundity datasets. After evaluation and comparison of this study model with other existing and baseline models, this study outperforms the others across the datasets used. The proposed encoder-LSTM model showcased remarkable advancements in accurate fecundity prediction, culminating in precise binary class predictions in time series data. This research's findings not only contribute to refining fecundity analysis but also signal a promising stride toward accurate predictions in healthcare applications reliant on time series data.</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

Enhancing Women's Fecundity Prediction in Time Series Data Using Encoder-LSTM Model Integration

  • 期刊论文

  • 发表日期：

    2025年7月14日

查看全文
  • Ahmad Shehu Muhammad,
  • Muhammad Bashir Abdullahi,
  • Mohammed Danlami Abdulmalik,
  • Opeyemi Aderike Abisoye

摘要

Recognizing the paramount importance of accurate fecundity prediction in assessing pregnancy possibilities, this research aimed to contribute significantly to healthcare methodologies by introducing a novel methodology in fecundity time series data analysis. While previous endeavors incorporated deep learning, notably LSTM models, integrating subfertility concepts, their limitations in handling extensive global data dependencies posed challenges to prediction accuracy. This study proposed a unique approach—fusing LSTM, termed as LSTMUE in the fecundity analysis domain, into a transformer encoder architecture with unique attention head selection mechanism. The evaluation of this proposed methodology utilized two distinct fecundity datasets. After evaluation and comparison of this study model with other existing and baseline models, this study outperforms the others across the datasets used. The proposed encoder-LSTM model showcased remarkable advancements in accurate fecundity prediction, culminating in precise binary class predictions in time series data. This research's findings not only contribute to refining fecundity analysis but also signal a promising stride toward accurate predictions in healthcare applications reliant on time series data.