What question did this study set out to answer?

This research aims to identify suitable habitat areas for Lyme disease and tick-borne encephalitis vectors in Japan using machine learning.

May 16, 2026Open Access

Detection of suitable habitat areas in Japan of the Lyme disease and tick-borne encephalitis vectors Ixodes ovatus and Ixodes persulcatus based on abiotic factors

Key Points

This research aims to identify suitable habitat areas for Lyme disease and tick-borne encephalitis vectors in Japan using machine learning.
Developed machine-learning models using tick surveillance and environmental data to predict habitats.
Collected data from 93,289 ticks across 57 sources for model training.
Utilized Random Forest algorithms based on 26 abiotic variables such as climate and topography.
Identified high suitability areas in Hokkaido and cooler, wetter regions of central Honshu for both Ixodes species.
Found I. ovatus had a broader ecological niche with moderate suitability in central Kyushu and Tokyo Bay area.
Machine learning algorithms showed high predictive performance with area under the curve > 0.89.

Abstract

In Japan, Lyme disease and tick-borne encephalitis (TBE) are primarily in the northernmost prefecture Hokkaido, where their primary vectors Ixodes ovatus and Ixodes persulcatus are most abundant. Recently, tick surveillance activities have collected both species across Japan, indicating potential expansion of their tick-borne pathogens (TBPs). We built a machine-learning (ML) model using available tick surveillance and environmental data to predict suitable habitat areas for I. ovatus and I. persulcatus and identify potential higher-risk areas of Lyme disease and TBE across Japan. Data on the occurrence and abundance of 11 vector tick species between 1990–2023 were extracted from studies identified via systematic literature search in two online databases or provided by local experts for ML development. After multiple iterations and permutations, separate Random Forest ML algorithms for I. ovatus and I. persulcatus were trained via 26 abiotic variables of climate and topography based on the occurrence and abundance respective to each species. Data on 93,289 ticks from 57 sources were extracted, and the ML algorithms’ area under the curves were high (> 0.89). Climate-related variables were the strongest predictors (> 90% cumulative model importance) for both I. ovatus and I. persulcatus . High suitability for both species was identified in Hokkaido and cooler, wetter regions in central Honshu, while I . ovatus had a broader ecological niche than I. persulcatus , with moderate suitability in mountainous regions of central Kyushu and surrounding the Tokyo Bay area. Our ML models suggest high suitability areas for Ixodes vectors may be widespread in Japan, indicating expanding potential risks of TBPs. • A dataset with 93,289 ticks on 11 species in Japan was compiled and provided. • Suitable areas for Ixodes vectors are predicted throughout Japan. • Tick-borne diseases by Ixodes ticks may be more widespread in Japan. • Machine-learning is A useful approach to inform potential tick-borne risks.

Read Full Paperexternally

Perguntar à IA

Bookmark

View Full Paper