In this article we discuss a research paper by Alcantaraa et al  named ” Improving tuberculosis diagnostics using deep learning and mobile health technologies among resource-poor communities in Perú ” in which Artificial Intelligence is used for medical applications. In this research published in 2017, highly infectious disease Tuberculosis (T.B.) has been targeted to be diagnosed well, by proper AI and Mobile Computing help. As per research paper  in 2015 10.4 million people around the world were infected by T.B.. It has both an application of mobile computing and artificial intelligence, in particular, deep learning networks. The aim is to timely detect TB so as to stop the epidemic form rising in effected places.
The key steps discussed here are:
- Firstly, data, a machine can learn given the data, the data which is tagged as coming from T.B. patients is required.
- The key constrained is the place where it has to be studied, places having deficient resources to do mass level expensive diagnosis and analysis.
- The X-Ray data is collected via mobile devises
- The data is tagged with help of doctors. Scientists from USA and Peru worked on this and created a database for TB images.
- The data is then classified as of different severity by annotators, images were provided with interfaces to classify the severity. Some evident categories include polygons in X-Ray images.
- This is fed into AI classifier which is taken as convolution neural networks for this research by researchers.
- And new data is classified once a deep learning model is learned
The analysis was performed in places such as Peru. The key contributions of paper  are
- Provided large database of X-Ray images for detecting the stage of TB and the database is verified by Pulmonologist.
- Building up Artificial Intelligence based model for detecting and classifying a patient as a TB patient and giving the required health care facilities on time to save lives
Comments: This huge database can be used to test around the world and even can be compared to the X-Ray images in current scenarios. This is a great research work, which involved tagged database creation and also the classification of a new person X-Ray as a TB patient or not. I hope they are using the build models in various hospitals to give an good ending to their research and hardwork. This is because the training data may not vary much and I suggest to the authors and Scientists to use these models and image databases in local hospitals around the world. This can also give an analysis of how other lung infections related to T.B. expert-annotated data. And how much it tally with the lung infections in Corona. Only thing is to provide the classification toolkit to local hospitals.
 Alcantaraa et al., “Improving tuberculosis diagnostics using deep learning and mobile health technologies among resource-poor communities in Perú”, Smart Health 1-2, pp. 66-76, 2017.