A deep learning based hybrid model for maternal health risk detection and multifaceted emotion analysis in social networks

In the field of public health, accurately identifying maternal health risks through social network data is both vital and challenging due to the complexities of multimodal sentiment analysis. Our study addresses this challenge by introducing the maternal health risk factor detection using deep learning approach (MHRFD-DLA), a novel framework that integrates convolutional neural networks, long short-term memory networks, and attention mechanisms. This approach enhances sentiment analysis and risk detection in maternal health, with the focus on critical areas such as prenatal care, mental health, and nutrition. MHRFD-DLA utilizes multimodal data, including text and electrocardiogram (ECG) signals, offering a comprehensive assessment of maternal health risks. Our model outperforms existing multimodal sentiment analysis models, achieving an accuracy of 98.4%, a precision of 97.6%, a recall of 95.6%, and an F1 score of 98.4%. Through performance evaluations, visualizations such as the confusion matrix and class distributions further validate its robustness. The MHRFD-DLA model not only bridges significant gaps in current methodologies, but it also sets a new benchmark for maternal health surveillance and intervention, demonstrating its practicality and effectiveness in real-world applications.

---

Opracowanie rekordu ze środków MNiSW, umowa nr POPUL/SP/0154/2024/02 w ramach programu "Społeczna odpowiedzialność nauki II" - moduł: Popularyzacja nauki (2025).