Prediction of dengue cases using the attention-based long short-term memory (LSTM) approach.

This research proposes a 'temporal attention' addition for long-short term memory (LSTM) models for dengue prediction. The number of monthly dengue cases was collected for each of five Malaysian states i.e. Selangor, Kelantan, Johor, Pulau Pinang, and Melaka from 2011 to 2016. Climatic, demographic, geographic and temporal attributes were used as covariates. The proposed LSTM models with temporal attention was compared with several benchmark models including a linear support vector machine (LSVM), a radial basis function support vector machine (RBFSVM), a decision tree (DT), a shallow neural network (SANN) and a deep neural network (D-ANN). In addition, experiments were conducted to analyze the impact of look-back settings on each model performance. The results showed that the attention LSTM (A-LSTM) model performed best, with the stacked, attention LSTM (SA-LSTM) one in second place. The LSTM and stacked LSTM (S-LSTM) models performed almost identically but with the accuracy improved by the attention mechanism was added. Indeed, they were both found to be superior to the benchmark models mentioned above. The best results were obtained when all attributes were included in the model. The four models (LSTM, S-LSTM, A-LSTM and SA-LSTM) were able to accurately predict dengue presence 1-6 months ahead. Our findings provide a more accurate dengue prediction model than previously used, with the prospect of also applying this approach in other geographic areas.

A Deep Learning Approach for Dengue Fever Prediction in Malaysia Using LSTM with Spatial Attention.

This research aims to predict dengue fever cases in Malaysia using machine learning techniques. A dataset consisting of weekly dengue cases at the state level in Malaysia from 2010 to 2016 was obtained from the Malaysia Open Data website and includes variables such as climate, geography, and demographics. Six different long short-term memory (LSTM) models were developed and compared for dengue prediction in Malaysia: LSTM, stacked LSTM (S-LSTM), LSTM with temporal attention (TA-LSTM), S-LSTM with temporal attention (STA-LSTM), LSTM with spatial attention (SA-LSTM), and S-LSTM with spatial attention (SSA-LSTM). The models were trained and evaluated on a dataset of monthly dengue cases in Malaysia from 2010 to 2016, with the task of predicting the number of dengue cases based on various climate, topographic, demographic, and land-use variables. The SSA-LSTM model, which used both stacked LSTM layers and spatial attention, performed the best, with an average root mean squared error (RMSE) of 3.17 across all lookback periods. When compared to three benchmark models (SVM, DT, ANN), the SSA-LSTM model had a significantly lower average RMSE. The SSA-LSTM model also performed well in different states in Malaysia, with RMSE values ranging from 2.91 to 4.55. When comparing temporal and spatial attention models, the spatial models generally performed better at predicting dengue cases. The SSA-LSTM model was also found to perform well at different prediction horizons, with the lowest RMSE at 4- and 5-month lookback periods. Overall, the results suggest that the SSA-LSTM model is effective at predicting dengue cases in Malaysia.

Leptospirosis modelling using hydrometeorological indices and random forest machine learning.

Leptospirosis is a zoonosis that has been linked to hydrometeorological variability. Hydrometeorological averages and extremes have been used before as drivers in the statistical prediction of disease. However, their importance and predictive capacity are still little known. In this study, the use of a random forest classifier was explored to analyze the relative importance of hydrometeorological indices in developing the leptospirosis model and to evaluate the performance of models based on the type of indices used, using case data from three districts in Kelantan, Malaysia, that experience annual monsoonal rainfall and flooding. First, hydrometeorological data including rainfall, streamflow, water level, relative humidity, and temperature were transformed into 164 weekly average and extreme indices in accordance with the Expert Team on Climate Change Detection and Indices (ETCCDI). Then, weekly case occurrences were classified into binary classes "high" and "low" based on an average threshold. Seventeen models based on "average," "extreme," and "mixed" indices were trained by optimizing the feature subsets based on the model computed mean decrease Gini (MDG) scores. The variable importance was assessed through cross-correlation analysis and the MDG score. The average and extreme models showed similar prediction accuracy ranges (61.5-76.1% and 72.3-77.0%) while the mixed models showed an improvement (71.7-82.6% prediction accuracy). An extreme model was the most sensitive while an average model was the most specific. The time lag associated with the driving indices agreed with the seasonality of the monsoon. The rainfall variable (extreme) was the most important in classifying the leptospirosis occurrence while streamflow was the least important despite showing higher correlations with leptospirosis.

Using satellite-measured relative humidity for prediction of Metisa plana's population in oil palm plantations: A comparative assessment of regression and artificial neural network models.

Metisa plana (Walker) is a leaf defoliating pest that is able to cause staggering economical losses to oil palm cultivation. Considering the economic devastation that the pest could bring, an early warning system to predict its outbreak is crucial. The state of art of satellite technologies are now able to derive environmental factors such as relative humidity (RH) that may influence pest population's fluctuations in rapid, harmless, and cost-effective manners. This study examined the relationship between the presence of Metisa plana at different time lags and remote sensing (RS) derived RH by using statistical and machine learning approaches. Metisa plana census data of cumulated larvae instar 1, 2, 3, and 4 were collected biweekly in 2014 and 2015 in an oil palm plantation in Muadzam Shah, Pahang, Malaysia. Relative humidity values derived from Moderate Resolution Imaging Spectroradiometer (MODIS) satellite images were apportioned to 6 time lags; 1 week (T1), 2 weeks (T2), 3 week (T3), 4 weeks (T4), 5 week (T5) and 6 weeks (T6) and paired with the respective census data. Pearson's correlation was carried out to analyse the relationship between Metisa plana and RH at different time lags. Regression analyses and artificial neural network (ANN) were also conducted to develop the best prediction model of Metisa plana's outbreak. The results showed relatively high correlations, positively or negatively, between the presences of Metisa plana with RH ranging from 0.46 to 0.99. ANN was found to be superior to regression models with the adjusted coefficient of determination (R2) between the actual and predicted Metisa plana values ranging from 0.06 to 0.57 versus 0.00 to 0.05. The analysis on the best time lags illustrated that the multiple time lags were more influential on the Metisa plana population than the individual time lags. The best Metisa plana prediction model was derived from T1, T2 and T3 multiple time lags modelled using the ANN algorithm with R2 value of 0.57, errors below 1.14 and accuracies above 93%. Based on the result of this study, the elucidation of Metisa plana's landscape ecology was possible with the utilization of RH as the predictor variable in consideration of the time lag effects of RH on the pest's population.