LungAI

As of June 2020, Covid-19 is everywhere and has more than 8 million cases reported Rate of spreading of COVID-19 and lack of disease specific symptoms has made manual detection more and more difficult.

LungAI aims to make Covid-19 detection accessible as well as reliable. It also makes Lung Cancer detection and detection of other common and dangerous Lung Abnormalities like Pneumonia, Edema etc. available so that users may not only diagnose whether they are positive for Covid-19 but also detect other medical issues in case they test negative for the virus.

We solve the issue of detecting coronavirus among populations by using dual detection techniques of analyzing chest image patterns and user inputted symptoms to give a final diagnosis.

LungAI aims to use diagnostics as a frontier for Covid-19 surveillance and speed it up so that the detection process can be made available without risking safety or reliability.

India has a population of over 1.3 billion people and also has very high population density. Rapid and large scale testing is the only way to prevent uncontrolled spreading of COVID-19. Yet due to lack of testing requirements in the form of testing kits, safety equipment etc manual detection in India is unable to detect cases faster than the spread.

And this is the biggest issue LungAI wishes to solve. The issue of physically having to test individuals for COVID-19 reduces efficiency and also puts the lives of testing teams in danger due to danger of transmission. Further, the fastest testing kits take upwards of an hour to give detection.

LungAI wishes to make available a single platform for users all around the world to use simultaneously for detection of coronavirus among large populations. In this manner if diagnostic procedures are sped up larger number of people can be diagnosed and protected by using necessary mitigating actions to reduce extent of damage caused by the virus to as large an extent as possible.

Our web platform makes use of radiographic or computed chest images and a secondary COVID-19 Risk Assessment Software to give detection of coronavirus within 3-5 seconds.

LungAI as a web application requires a photograph of the patient’s lung CT Scan or X-Ray to be uploaded. It tests this image against the Covid-19 detector and gives a probabilistic answer as to whether the user is positive for the coronavirus or not. In case the user tests negative they may then go further and test for lung cancer as well as other lung diseases.

This system works in tandum with LungAI’s Covid-19 Risk Assessment System which records a brief history of the users medical conditions as well as his current symptoms with regard to coronavirus and then gives its assessment in terms of the user being high, moderate or low risk with further instructions to mitigate effects of the virus to as large an extent as possible. The added benefit of the Software is that users can test their symptoms for Covid-19 in case they wish to but are unable to procure the required radiological or computed images.

Thus based on features extracted from the users chest X-Ray or CT Scan as well as their symptoms LungAI can then give a more accurate judgment about the patients current and future probability of having Covid-19.

As of June 2020 India has reported over 367,000 COVID-19 cases and over 12,000 deaths.

In terms of target population LungAI aims to begin by deploying the software in states of Delhi and Maharashtra, the two worst hit Indian states followed by deployment in eventually all states across India.

The web application being on the internet will always be available to people of all working classes and financial backgrounds provided they have access to internet, required images for testing or wish to only test their symptoms for coronavirus risk assessment.

LungAI will also be available in all COVID-19 test sites as these will be the areas of maximum testing stress. In this too the platform will first be deployed in Red Zone areas having highest COVID-19 density followed by orange zones and finally green zones which have not reported any new cases but might.

The web application has also been provided with an automatic medical contacting system in case the user develops symptoms or the application detects a presence of COVID-19. The application further also contains instructions for the user to contact the nearest COVID-19 test sites as well as measures to reduce immediate effect of virus in his surroundings.

LungAI endeavours to mitigate the effects of coronavirus to the best of its ability while proposing certain changes in various medical strategies. We wish to expand the definition of health and medical security to not just patients but also to medical staff and health workers who are currently risking their lives to save ours. LungAI aims to change this and provide a unified procedure where lives of all people are protected while providing a significant improvement using technology to current medical procedures. In this manner we wish to protect lives, but not only those of patients, rather everyone’s life.

1)LungAI provides support to both chest radiograph and computed tomography images for all the diagnostics tests it offers. It requires image of the patient’s lung CT Scan or X-Ray to be uploaded and tests it against the Covid-19 detector which gives a probabilistic answer as to whether the user is positive for coronavirus or not. It also provides detection of lung cancer and other lung abnormalities as well.

2)To further make the diagnostic process more accurate LungAI offers a Covid-19 Risk Assessment System which records brief history of the users medical conditions as well as his current symptoms with regard to coronavirus and then gives its assessment in terms of the user being high, moderate or low risk with further instructions to mitigate effects of the virus.

3)Unlike current testing kits our diagnostic technique is completely non invasive. As the entire process is completely digital, diagnosis requires no physical contact with patients thus reducing spread of the virus and promoting safety of medical-staff.

4)Being digital LungAI can be accessed at numerous locations simultaneously solving the problem of lack of testing kits.

5)As it may be difficult for certain users to completely trust an App for correct diagnosis, the application has been provided the ability to send relevant images along with the application’s diagnosis to verified doctors who can then give the user a final judgment based on all the data.

6)Also while the fastest testing kits take upwards of an hour, LungAI can give its detection within about 3 seconds.

1) LungAI as the name suggest uses a form of artificial intelligence that finds patterns among images of a particular subtype.

2) Convolutional Neural Networks (CNN).   For the purpose of this project, I made use of Convolutional Neural Networks, a type of Deep Neural Network, most commonly applied to analyse visual imagery. Convolutional Neural Networks are different from normal neural networks because they contain a special type of layer called a Convolutional Layer, which contains a filter that is able to understand certain types of patterns in the image.

3) To build the CNN I used native Keras which is an open-source neural-network library written in Python. I used Transfer Learning to create this model, which means that I took a pre-trained network called MobileNet which is trained with ImageNet, a dataset of over 14 million images, and added more layers to it so that it could classify various lung conditions including Covid-19, lung cancer and other lung based diseases. The idea is that we take a model that was trained with a huge amount of data and use that for a new task where we do not have as much data. We basically take a pre-trained model that can detect images and add a few more restrictions to the images being detected so that it can detect factors we need and ignore the ones we do not need. Embedding of the model onto the website has been done using JavaScript.

LungAI Main Website: www.lungai.in

1) The use of artificial intelligence and especially the use of convolutional neural networks in the field of medical diagnostics and imaging though rather new has gained widespread popularity in a very short period.

2) While LungAI is one of the first tech-startups to use Chest Radiograph and Computed Tomography images for detection of COVID-19[1][2] the use of such medical images has previously been done in case of detection of lung cancer as well as detection of other lung diseases like Pneumonia[3].

3) Specifically the use of Transfer Learning has played a huge role in not just Chest X-Ray/CT Scan based diagnostics but has also been used in other medical procedures like Skin Cancer detection[4].

4) But what LungAI[5] does to set itself apart from the rest is the fact that it provides users with the benefit of getting themselves tested for all possible lung abnormalities, be it lung cancer or some other disease like Pneumonia or Tuberculosis[6]. In case the user tests positive for Covid-19 the user may then take necessary action to protect himself and his surroundings but the website also has an option for patients other than those affected by coronavirus.

6) Apart from this the website further gives a two stage detection by asking users questions regarding their medical history as well so as to give a complete judgement of the users medical condition based on not only the image uploaded but also the users current symptoms.

7) Product Demo : 

1) In terms of verification that the solution proposed by LungAI works we have achieved an overall accuracy of over 98% when the website is run using only the COVID-19 Radiological Detector and achieved a very high accuracy of 99.67% when the Radiological Detector was used in sync with the Risk Assessment Detector requiring both chest radiograph/computed tomography image as well as the users symptoms.

2) These figures have been verified by the government and its respective authorities on a dataset made available by them. We have further also approached the Indian Council of Medical Research(ICMR) to validate the given solution on their own database of chest images and if approved collaborate with the Indian Government to fully deploy this web application on a much larger scale.

3) Apart from this in terms of certain preliminary tests that we had run, among 50 randomly chosen patients the model detected 47 of the patients correctly. Of the three misdiagnosed, 2 were suffering from cases of acute and mild pneumonia while the third user had mild consolidation and difficulty in breathing through his lungs.

4) Further plans for verification and validation as well as deployment include use of the software in 5 hospitals in Delhi, which happens to be one of the worst affected states by COVID-19. Following the preliminary testing in these hospitals LungAI will then be given the green light on whether it can fully deploy in 50 test sites and hospitals.

5) We feel LungAI is a viable testing procedure for COVID-19 because it solves some of the many major issues provided by the current testing kits and methods, namely efficiency and speed as well as safety of health workers and staff. It would also further solve the issue of testing kits being unavailable as the website can be accessed by multiple users simultaneously without any limit. The proposed solution is also cost effective in the manner that is does not require any specialised equipment to function thus immensely increasing its accessibility.

Further we would also be interested in extending our research to public health authorities for further development. 

1)LungAI is currently undergoing prototype testing under various government agencies and has only been operational for 2 months. In this time LungAI has run over 275 tests as part of its verification on live patients. In terms of the number of X-Rays LungAI has tested so far, we recently crossed 10000 tests.

2)In case LungAI is given the green light in terms of accuracy and functioning by Health Ministry and Indian Council of Medical Research (ICMR) we plan to begin operations in 5 hospitals in Delhi followed by 50 hospitals in 6 months with full scale operation in all COVID-19 test sites where we will be deployed.

3)In terms of testing capacity within the first one year the number of tests conducted largely depends on the imaging capabilities but being software and estimating current conditions LungAI has the capability of running over 50000 tests daily per web application accessed. But being a digital solution, the more the available devices that higher the testing rate can be. In terms of our preliminary test, it was found that the software could make over 5000 detections within a matter of a 2-3 minutes provided availability of X-Ray or CT Scans is taken care of.

4)But scope of daily detections that can be made are unlimited given that the website can make over 50000-60000 detections per tester in a given hospital provided patients have chest images readily available. Over the next 5 years we aim to fully deploy in over 500-600 hospitals in India.

1)Following preliminary testing and validation of the software LungAI will initially be given support in deploying our web application in 5 hospitals in Delhi under normal testing conditions. This will be done to see and improve the overall functioning of the software under normal testing stress to improve efficiency.

2)In terms of our one year goals we aim to fully deploy in at least 75-80 hospitals across India with a majority of our testing sites being present in Red Zone particularly in the 5 worst hit states of India. This will be followed by complete operational support by the Health Ministry to further integrate our system in other medical facilities around India for further testing of the software.

3)We also aim to ramp up our testing capabilities by recruiting testers to fully control and operate the software as more testers imply more systems used which would improve number of tests conducted daily and will eventually help mitigate effects of the virus better.

4)In terms of our goal and aims within the next 5 years we aim to be operation in over 500-600 hospitals in India and if possible also partner with other international medical research teams to further our solution to other nations for use. In terms of the next 5 years we also don’t expect only COVID-19 testing to remain our main operation and also expect an increase in other tests we provide such as that of lung cancer detection and will be working towards bettering those resources.

Artificial intelligence based application have only one shortcoming and that is the availability of readily available and verified data to prepare the software on. While the current solution has very high accuracy more verified data is always better and can help further the robustness and accuracy of the software.

 The other major issue we most medical technologies face are the implication caused due to long and many validation procedures before real-time deployment in medical test sites. 

Apart from this financial backing is also a major issue for LungAI as we cannot function without availability of required resources and workforce. Apart from this finding investors to give us financial backing is not easy or readily available.

1)In terms of overcoming the lack of available datasets we had already predicted the issue and hence had proactively requested the government and other reputed research institutes for data which we duly received and were thus able to conduct and develop our software.

2)We also had started our medical testing procedures very early on in development and thus had to wait very little time between validation and verification and actual large scale deployment.

3)In terms of funding so far our operations so far have been largely funded by a combination of grants and investment capital. We would like to mainly subsidise the service provision of testing. Making testing accessible to everyone and keeping our software alive with the proceedings is currently our main target.

4)But in the eventuality that we run of services we will likely have to impose certain costs per user per test administered by the web application. These costs will further be used to operate the required Research and Development while maintaining the diagnostics process of LungAI.

5)Apart from this in case the above platform gets certified for use by the government of India and the Health Ministry we would also be hopeful of a partnership with the involved government agencies for large scale deployment of LungAI.

We are also completely open to investments and grants available as well as other resources which could help further our research and improve it or make it further deployable in other areas.

While we are currently not registered as an organization we have filed for a patent for the same through the Indian Patent Office and will be registering ourselves as a organization as soon as we have deployed in Delhi and have government validation and verification.

LungAI was initially an idea that I had developed to provide healthcare to everyone without then having to ever visit the hospital or medical clinic for testing. I had aimed to provide an in house doctor to all users through the power of technology without them having to ever meet anyone.

Yet I would never have come so far had it not been for the LungAI team.

We have limited staff including 5 full time (CEO, CTO, Software Developer/Backend Developer and Communications Head) and 6 part time(Backend Developers, Customer Services, Design) staff working on our current products.

The LungAI team is mix of various members of various technological and other spheres that have the ability to synergise their skills productively to make any product a working success and our current solution, the Digital COVID-19 detector is no exception to this rule.

LungAI is a culmination of all of our efforts over the past 5-6 moths and has so far proved to be a success and is on its way to becoming India’s first digital COVID-19 detector to be largely deployed on a real scale.

Our team consists of members largely with a technical background with almost half our staff being involved in the backend and software of LungAI’s web application which is also precisely one of the main reasons contributing to the large scale acceptance and validation of LungAI’s COVID-19 solution.

The rest of the members played an instrumental role in the frontend design of the application and most importantly the user experience of the application. The application has thus been designed with as clean of colors and textures as possible so as to help users focus on our products rather than the design it-self.

Other than that the team is a highly motivated and dynamic set of people who are ready to pool in their skills into what has otherwise has been a really tough and difficult but rewarding journey so far.

The application has so far passed all the validation and verification requirements and is slated to be deployed around July end in 5 hospitals.

Currently LungAI has had no partnerships or collaboration with any other organizations except in terms of requesting information and verified datasets which was largely useful for research conducted by the LungAI team. As of now our current collaborators are only in the form of various health ministries and the government largely in terms of initial deployment and verification of our products functioning.

Though in the future we are openly ready for collaboration in terms of research and deployment not just in India but also overseas. Apart from this we are also open to research prospects with other medical imaging teams as well as other pharmaceutical companies to further research not just for COVID-19 research but also in other fields of medicine and biotechnology.

We have to some extent collaborated with John Hopkins University, 

Italian Society of Medical and Interventional Radiology(SIRM), 

The Lung Image Database Consortium image collection (LIDC-IDRI) and the National Institute of Health(NIH) but only in terms of research and accessibility of verified data and information regarding the current pandemic and other lung based abnormalities faced by patients.

LungAI serves a dual aim in the sense that while at its core it is a dual level diagnostic procedure for diagnosis and detection of COVID-19 it can also be used as a large scale surveillance software to further sift out the areas where testing of COVID-19 needs to be carried out. In this manner it can also used as an application to find areas with high COVID-19 density which requires isolation and other medical facilities.

On the other hand it serves as a reliable and quicker diagnostics procedure to speed up the detection frontier of coronavirus. Apart from this the software can also be used as part of a large detection procedure wherein it can be deployed for radiologists in hospitals to further rapid testing methods in hospitals.

LungAI currently serves limited products in terms of coronavirus security in the form of a Radiological/Computed Tomography Image Detector as well as a COVID-19 Risk Assessment Software which work in tandum to give users a final judgement on their current probability of having coronavirus. Apart from this the software also provides users and hospitals the ability to detect lung cancer as well as other lung abnormalities like Pneumonia, Tuberculosis, Edema etc.

Requirement is huge in current times where spread of the virus is at its worst, worsened by lack of testing kits, safety equipment as well as probability of false negatives of the RT-PCR tests. This also comes at a time when the lives of medical staff are also under danger.

Currently LungAI has not yet signed any contracts with any other organizations for investments and has largely maintained finances through crowd-funding options and support from the government and health ministry with respect to deploying our software in their hospitals.

While sustained donations could keep us going for another 7-8 months with our research and development team fully functional we are currently in talks with mainly the government for future funding and investment options.

Our main source of income comes from users in the form of costs paid for every 2 tests , mainly for users to avail of an X-Ray based test as well as a CT Scan based COVID-19 test. The Risk Assessment Software on the other hand is free of cost for users to avail of as many times as they wish to.

Ideally we would like to further lower our costs due to requirement of large scale testing of COVID-19 in India specially for the general population and thus we are currently in talks with various organization for funding and investment opportunities.

Apart from that we are also looking into collaboration with various private hospitals and pharmaceutical agencies to further COVID-19 research and development in collaboration with the LungAI team along with investment ventures in LungAI.

In terms of support LungAI is applying to Solve with the main aim of finding funding opportunities as well that of collaborating with other like minded companies which share similar ethics and principles. 

In that LungAI aims to further its research capabilities with other organizations to further medical research.

We wish to partner with organizations at an international level so as to collaborate our research and development capabilities with theirs to further our product as well as improve our services. Further collaboration would benefit various organizations in terms of increased work force provided by both companies.

We would like to partner with MIT itself as they are one of the leading research institutes in the world and LungAI feels it would benefit the most by this initiative if formed.Other than that we would be interested in collaborations with pharmaceutical companies that are heavily into COVID-19 vaccine development like Pfizer etc to further COVID-19 research all over the world.

One of the most important aims of LungAI has been to further the safety of medical staff which is often overlooked and we do so by providing a completely contact less and safe diagnostics process.

We will further use the funds to expand our network of LungAI supported health centers and aim to further lower our testing costs to as low and affordable a price as possible. Further we will invest in other medical fields other than corona virus detection and spread our domain to other medical fields.

LungAI is channeling its software and backend development skills into mitigating the effects of the coronavirus to as large an extent as possible and thus aims to further the safety of both patients as well as medical staff against the deadly pandemic.

We will further use the funds to expand our network of LungAI supported health centers and aim to further lower our testing costs to as low and affordable a price as possible. Further we will invest in other medical fields other than corona virus detection and spread our domain to other medical fields.