Authorship：Lead author   Publishing type：Research paper (scientific journal)   Publisher：Springer Science and Business Media LLC  

Abstract

This study aims to evaluate the effect of latent diffusion models on molecular representation learning from the perspective of generalization performance in molecular property prediction. To this end, we formulate a deep generative model for molecular representation learning based on a latent diffusion–based prior distribution, and introduce an evaluation methodology of generalization for learned molecular representations using the widely applicable information criterion (WAIC) and the widely applicable Bayesian information criterion (WBIC). Furthermore, we propose an analysis framework based on smoothness and multi-modality to analyze the factor of generalization in molecular representations. We constructed the graph latent diffusion autoencoder (Graph LDA), a deep molecular generative model that combines a transformer-based graph variational autoencoder and latent-diffusion-based latent prior distribution, designed to construct graph-level molecular representations through unsupervised learning. We compared the generalization performance of Graph LDA with other molecular representation learning models using WBIC and WAIC across multiple molecular properties, including HOMO energy, solubility, and biological activities. The results demonstrate that molecular representations learned by different models exhibit distinct generalization behaviors, and that representations learned by Graph LDA—using a latent diffusion–based prior—consistently show improved generalization in molecular property prediction. Using our proposed framework, we empirically demonstrate that the superior generalization performance of Graph LDA is attributable to the smoothness and multimodality of its learned molecular latent representation. These findings provide a principled understanding of the role of latent diffusion–based molecular representation learning in improving generalization performance.

Scientific contribution : This work systematically analyzed the effect of latent diffusion–based priors in molecular representation learning from the perspective of generalization performance in molecular property prediction. Through generalization evaluation using WBIC and WAIC, together with an analysis framework for molecular representations, it is empirically demonstrated that latent diffusion–based priors contribute to deep generative models extracting smooth and multimodal latent representations, which in turn lead to enhanced generalization performance of molecular representations. These findings offer a principled guideline for developing molecular representation learning models with high generalization.

DOI： 10.1186/s13321-026-01176-8

researchmap

Authorship：Lead author   Publishing type：Research paper (scientific journal)   Publisher：Frontiers Media SA  

Background

Influenza remains a significant public health challenge worldwide, necessitating robust forecasting models to facilitate timely interventions and resource allocation. The aim of this study was to develop a long short-term memory (LSTM)-based short-term forecasting model to accurately predict weekly influenza case counts in Tokyo, Japan.

Method

By using weekly time-series data on influenza incidence in Tokyo from 2000 to 2019, along with meteorological variables, we developed four distinct models to evaluate the impact of the external variables of mean temperature, relative humidity, and national public holidays. After model training, we assessed the predictive performance on an independent test dataset, using mean square error (MSE), root mean square error (RMSE), mean absolute error (MAE), and Pearson’s correlation coefficient.

Results

During the study period, 1,445,944 influenza cases were analyzed. The model incorporating all three external variables demonstrated superior predictive accuracy, with an MSE of 3,646,084, RMSE of 1,909, MAE of 849, and Pearson’s correlation coefficient of 0.924. These findings underscore the substantial contribution of these external factors to improving the prediction performance.

Conclusion

This study highlighted the efficacy of LSTM-based models for short-term influenza forecasting and reinforces the importance of integrating meteorological variables and national public holidays into predictive frameworks. Our optimal model provided more precise forecasts of influenza activity in Tokyo, Japan.

DOI： 10.3389/fpubh.2025.1618508

researchmap

Authorship：Lead author  

File： 2312.13110.pdf

researchmap

Authorship：Lead author  

File： GraphLDA.pdf

researchmap

Authorship：Lead author   Publishing type：Research paper (scientific journal)   Publisher：Wiley  

DOI： 10.1002/minf.202000203

researchmap

Other Link： https://onlinelibrary.wiley.com/doi/full-xml/10.1002/minf.202000203

Total pages：3, 6, 244, 6, 図版26p   Language：Japanese

CiNii Books

researchmap

Event date： 2024.10

Presentation type：Poster presentation  

researchmap

Presentation type：Poster presentation  

researchmap

Presentation type：Poster presentation  

researchmap

researchmap

researchmap

researchmap

共同研究：数理モデルと機械学習を活用した呼吸器ウイルス感染症の流行予測法の開発

researchmap

修士課程分

researchmap