How can AI warn about disease outbreaks?
Few of the best uses of artificial intelligence can be seen in potential fatal and time-sensitive conditions. The same is the case when a strange disease rears its head for the first time, affecting every aspect of life including businesses and governmental activities. It might appear difficult for government and public health authorities to perform a quick run across huge number of data within a short span of time and coordinate a response. Here comes the importance of state-of-the-art artificial intelligence technology in automatically gathering and mining through a pool of news reports and online content around the world. The technology can assist healthcare experts in figuring out the anomalies that lead to an epidemic or a pandemic.
BlueDot, a Canadian artificial intelligence firm, was in the spotlight for some weeks for warning the world about the novel coronavirus days prior to the official alarm from organizations like the Centres for Disease Control and Prevention (CDC) and World Health Organization (WHO). It used AI algorithm with full swing to process more data,
bring information available in different languages, collected reports from animal and plant disease tracking networks and airline ticketing data.
Data is the key
Conventional methods to analyse a disease outbreak is employed through studying the location the outbreak, the count of disease cases reported, the interval at which each cases are reported, from which parts of the world etc. can help forecast the possibility of the spread of the disease. As there are huge number of data sources available in a digital world, more information can be associated with the knowledge over the disease and make quick predications. One of the important problems with this way of analysis is most unstructured data such as news stories, blog write-ups and social media posts make it difficult for the AI algorithm to understand. Whereas, structured data with figures and location details can be tabulated efficiently, making it further easy to interpret.
Use of Deep Learning
For the disorientation created by unstructured data, deep learning techniques can help them in making sense. Utilizing artificial neural networks, the machine will be able to evaluate data with the help of processors arranged in each layer. Data can be evaluated and then transferred to the successive layers. It helps machines in recognizing the components of particular types of items. Since AI model’s capabilities are only good when the data is useful, it becomes critical in maintaining the quality too. The unstructured data might contain crowd-sourced data making it a challenging task to filter out unnecessary ones. This can be achieved by another method of verifying the result of the AI models. Incorrect prediction about a mass hysteria could lead to dangerous outcomes as a result of the spread of disease.
“The official information from verified authorities is not always timely,” says Kamran Khan, infectious disease physician and BlueDot’s Founder &CEO states in an interview. “The time difference that takes between one case of a person and the outbreak of a disease actually depends on how efficiently a healthcare worker recognizes the spread. That interval could be enough in preventing the outbreak from occurring at all,” adds Khan.