The last few months have been challenging for the medical fraternity who has seen one of their most demanding and difficult periods due to COVID-19. They have worked 24×7 in hospitals and other makeshift arrangements that sometimes do not have sufficient facilities to treat the COVID-19 positive patients. At such difficult times, technologies are also helping doctors to expedite Covid-19 diagnosis and treatment by helping them make the correct decision. This is the story of the doctors in a southern state in India who are leveraging Artificial Intelligence, WhatsApp group of village doctors for triaging their patients based on their chest X-rays.
Challenge of Rural Villages
Doctors in rural village areas often face the shortage of medical facilities and hence for major treatments, they have to send patients to district hospitals. This is a major challenge in current COVID-19 times as well. With limited medical facilities, doctors in rural areas and remote locations have to make right and timely decisions for COVID-19 patients and send them to district or city hospitals for better treatment.
This is where a group of WhatsApp community of doctor created by a doctor in Karnataka is helping them take these decisions with the insights of Artificial Intelligence.
WhatsApp Group and Machine Learning
A few months ago, Padmanabha Kamath a cardiologist in Mangaluru, Karnataka started a WhatsApp group called CAD – (Cardiology at Doorstep) in which he formed a community of doctors who would share ECGs of their patients within the group and discuss the best possible diagnosis and treatment.
The same group is now being leveraged by doctors from some Karnataka villages and districts to share the X-rays of COVID-19 patients.
It is to be noted that COVID-19 is known to affect the lungs. Although the RT-PCR (reverse transcription-polymerase chain reaction) test is the gold standard for the diagnosis of COVID-19 infection as on date, a chest X-ray is done to find out the involvement of lungs in COVID-19. Chest X-ray is beneficial for many patients who are critically ill and cannot wait for definitive reports for the RT-PCR test to come in.
Doctors in Whatsapp group have added to their decision-making the power of machine learning predictive models to identify the severeness of COVID-19. The machine learning model checks the chest X-ray of patients sent over WhatsApp to detect the presence and extent of COVID-19.
The level of infection in the lungs can often determine how the case can be treated. The intention behind this initiative is to decide if it is needed to recommend patients of village doctors to go to the district hospital for further treatment based on the results they get from their ML models and chest X-ray.
This initiative SWIFT (Simple WhatsApp-based Image interpretation at your Finger Tips) Rapid AI solution group has 125 doctors from various villages and towns in Karnataka state.
There are two ML models developed by IISc Bengaluru along with a start-up called NIRAMAI Health Analytics is crowdsourcing the photos of these X-rays from villages and districts for training the ML model. Both these models are trained in a different manner to analyze the chest X-ray and the results from both the models are given to the doctor concerned to help the take ensemble decision.
Dr. Padmanabha Kamath told The Hindu Business Line, “This sort of technology-driven triaging of patients helps since we know who the patients at risk are”. The doctors who are using this claim that this could be a breakthrough in the way we manage and handle this pandemic.
Geetha Manjunath, Chief Executive Officer of NIRAMAI Health Analytics that created these machine learning models claim that after RT-PCR results come in many doctors have given feedback that their models have predicted true positive case of COVID-19 with good accuracies.
The doctors who are a part of this group however do add a point saying that the outcome of the model prediction is indicative of the infection level and cannot be considered as a replacement to the RT-PCR test which is a must to determine whether a patient is COVID-19 positive or otherwise.
Questions are naturally being raised about data security and patient privacy. However, the medical fraternity is already careful about the handling of patient data and they take care of making data anonymous for the AI models.
We can surely say that two things are being achieved by this. We have a new way of tackling the COVID-19 pandemic and we are also getting a lot of crowdsourced data which is helpful to train machine learning models further.
In conclusion, village doctors looking for a proper result can make the right decision before getting the throat swab results. It helps in early detection and better immediate treatment as far as possible.