What model is used to predict dengue fever?
Dengue fever is certainly a common disease in tropical countries such as Malaysia. Beside Malaysia as one of the countries in the Southeast Asian region that have dengue cases, other countries such as in Africa and part of the America also have dengue cases. In Malaysia, dengue tends to peak around May to September during the dry southwest and November to March of rainy northeast monsoon seasons. This year, which is 2023, is predicted to be the peak of a major outbreak for dengue fever.
Dengue fever is caused by the mosquito bites of the Aedes mosquito. There’s a catch to this. Not all Aedes mosquitoes are able to infect humans with the dengue virus. Only the female Aedes mosquito is able to spread dengue virus to humans. In addition to this, only infected Aedes mosquitoes are able to do this. An Aedes mosquito is said to be infected when it bites a human with dengue virus circulating in the human blood. When the mosquito bites and feeds off the blood infested with the dengue virus, the mosquito is now infected with dengue virus. Thus, when this infected mosquito bites other humans, it releases the dengue virus into the human body. The person now is deemed to be infected with dengue.
The way how dengue fever is caused and spread explains why dengue fever is a transmittable disease. Furthermore, when the mosquito is infected, it has the capability of carrying the virus and continues to transmit the virus to humans for the rest of the mosquito’s life. This is why dengue is known as the fastest spreading viral disease through mosquito-borne diseases.
Do you know that 80{9cff5d18b6d4b9243ff4632f67e0ba40edeb1186cbe327dd41d5a4f25fbc7352} of the people infected by dengue do not have any symptoms or only have mild infections? Common symptoms include high-grade fever for more than 2 to 7 days and other accompanying symptoms such as headaches with pain behind the eyes (retro orbital headache), joint pain (arthralgia), muscle pain (myalgia) and skin rash. Treatment for dengue fever aims to support patients but in no means to eliminate the dengue virus from the body as there is no cure for dengue fever. Treatment such as fluid replacement therapy are the common treatment given for patients with dengue infection. Medicine such as paracetamol do help to relieve symptoms such as fever, headache and muscle or joint pain.
Since dengue fever is actually a transmittable disease, just as with many other transmittable diseases, knowing how to cope with it can help to reduce the severity of dengue fever and eventually reduce the number of those affected. One of the ways is to use a model that can predict dengue fever. Thus, what is the model used for such a purpose?
The most common model used to predict dengue fever is the long short term memory (LSTM). LSTM is a type of recurrent neural network that is well suited for modelling time series data such as data of the incidence of dengue fever over time. LSTM networks can learn long-term dependencies in data by using gates that control the flow of information within the network. This enables the LSTM network to detect patterns that may span multiple time steps. Such a network is useful in predicting future outcomes of dengue fever.
There are 2 types of attention mechanisms used by the LTSM network for dengue fever prediction. The first one is spatial attention which is a mechanism that allows models to weigh different features differently when making predictions. The second one is temporal attention which is suitable for dengue forecasting tasks. In Malaysia, the stacked LSTM with spatial attention (SSA-LSTM) model is effective in predicting dengue cases.
Predictive models are very useful in predicting dengue fever as it can help give information and serve as a warning for authorities to start taking steps. This includes targeted prevention measures such as mosquito control efforts and reducing impact of the outbreaks to the affected communities. Predictive model also helps to identify areas with resources that may be needed in the event of a dengue fever outbreak which further help public health officials to allocate resources in the targeted area. Last but not least, predictive models aid in identifying the patterns that give information on trends and changes in the risk of the outbreaks. This will help in refining the development of prevention and control strategies.
Dengue fever is a mosquito-borne disease that affects billions of people worldwide. Dengue remains endemic in Malaysia since its outbreak in the 1980’s with the highest recorded cases in the state of Selangor. Predictors of dengue fever outbreaks can provide timely information for health officials to implement preventative measures. This is very useful as symptoms of dengue fever may not reflect the true representation of the disease. There are more than one model used by the health authority to predict dengue fever but the common one and the one that seems to be the most reliable one in Malaysia is the LTSM model, specifically SSA-LSTM model.
