Khoj: Eradicating Tuberculosis and Redefining Epidemic Management

A novel AI based TB and drug resistant TB diagnosis device and epidemiological model aimed at monitoring, improving goal aligned metrics and designing strategies for TB Control. Enabling large-scale point of care testing, early treatment and focused management of Tuberculosis in India and the global south. 

Arastu Sharma, Researcher in optical computing, Centre for Photonics, University of Cambridge. Experienced in device development and prototyping. 

With more than 10 million cases detected every year, a million deaths and an increasing share of multi-drug resistant (MDR) strains, Tuberculosis (TB) pose an unignorable risk to global health. More than 20% of global population is estimated to have latent TB, which can become active disease during periods of low immunity (e.g. with HIV).  Rising MDR-TB prevalence at more than 500,000 `recorded` cases in 2019, poses a significant risk of spill over. Reduced access to diagnostics and treatment due to the COVID pandemic, may have led to half million+ excess deaths in 2020.

At recent rates, whilst writing this application, we estimate that nearly 500,000 persons would have died due to TB this year. TB is a disease of poverty as it spreads through vulnerable sections of the society.[9] Public health planners lack prevalence data, largely depending on passive case finding and screening. New point of care device technologies, such as ours, are required to increase accessibility for diagnosis and can provide prevalence data. Estimating, monitoring and improving on goal aligned metrics such as time taken to detect cases after symptoms appear and % of cases missed, is the need of the hour. Prevalence studies and modelling are required to better inform public health policy.

We are targeting the yearly burden of 50 million plus TB cases in the globe and increasing burden of drug resistant TB cases in the global south, starting with India. We aim at mitigating TB and eradicating MDR TB in local populations.

Public health agencies of developing countries have worked to mitigate the crisis of Tuberculosis for nearly a century. However, the burden of MDR-Tuberculosis remains very high in India and globally, projected to lead to 70 million deaths and a financial burden of 50 trillion dollars by 2050. The spill over risk of MDR-TB widely can be mitigated through timely discovery, treatment, isolation and appropriate healthcare practices. Often lack of information and data is the biggest barrier to policy and epidemiological interventions.[14] As an example, randomized TB diagnostic surveillance at the national scale occurred last in 1970s in India. Our cheap and portable technology, whose deployment may be guided by epidemiological considerations can enable such surveys to occur more frequently.

An end-to-end solution in the context of TB such as ours would be crucial to generate and analyze large data sets for active and passive cases. Monitoring the right metrics is key to identifying and solving the problem. We are collaborating multiple agencies – notably CSIR-Institute of Genomics (for device technology), LMJP Hospital Mumbai (for medical inputs), AIIMS- Jodhpur (for public health inputs), Department of Biotechnology, Rajasthan (for a pilot project), Avanti Finance (for manufacturing) to pilot our solution on the ground.

We plan to develop into a non-profit organization that supports governments and other agencies in fighting infectious diseases using our framework. Moreover, the solution is a non-profit initiative that is based on the ideals of solving infectious disease crisis using research, data generation and analysis Our solution is aimed at reducing TB and eradication of MDR and XDR TB in the world’s worst hit regions using a novel, portable, low-cost detection technology combined with our proven epidemiological modelling approach. The solution covers both ends from generation of large-scale data from local bodies to monitor the spread as well as using a large-scale data set to develop and propose our own epidemic management solutions. We will define goal-aligned metrics for understanding impact of the targeted diseases and create strategies to fight the spread. The generated data will be available for public research and policy planning as well as we plan to make the epidemiological model open source for further improvements, academic research and collaborations. Our goal is to develop an epidemiological framework for fighting infectious diseases, thus targeting and delivering results in the correct parameters. Also, to make the implementation of our idea successful on a large scale, we plan on delivering the diagnosis instrument at manufacturing cost.

Marco Schito et al, “Tuberculosis eradication versus control” have made a significant conclusion and have shown the importance epidemiological and social interventions to spread infection.[11] They connect this to an access to a large-scale extensive data, that can only be enables by a point-of-care device. Rukmini Shrinivasan et al “India’s syndemic of tuberculosis and COVID-19", makes case for active case finding with TB.[9] MDR-TB is a major threat to global health, with increasing potential of leakage into a uncurable drug-resistant microbe driven pandemic. The prevalence and increase in MDR-TB are predicted to be because of irregular treatment and epidemic mismanagement. Local government can greatly benefit and be dependent on data generated in understanding the infection spread. Our solution provides an alternative to enable this by a point of care testing diagnosis, with high sensitivity.

Mass testing, upon the timely advice from our team, solved the Covid problem in Krala, Andhra Pradesh, Tamil Nadu and Karnataka during the first wave of the pandemic. However, TB is more widespread social issue and we have come to understand the importance of long-term public health measures for this disease which we plan to bolster by increasing both testing and surveillance capabilities.

Year 1: We have finished the proof-of-concept stage for our artificial intelligence-based diagnosis device instrument, a revolutionary low-cost TB diagnosis test. We target prototyping the device and patenting the technology in the first six months, while providing verification on its efficacy with help of 1000 subjects.  Within the first year, research plans to setting up pilot projects in selected districts to corroborate the intervention planning will also be prepared.

Year 2: At this stage we plan on conducted a pilot projects involving 10k people with large scale active testing. Control group-based intervention planting, epidemiological modelling and social studies will also be conducted with the help of selected 250 public health centers and public health agencies. In parallel, after proof of concept runs for the overall impact structure – we plan to publish open source research in infection management.

Year 3: Will see expansion of the project across 2 Indian states, 50 districts and 2000 public health centers. One million people will be involved in the large-scale testing and screening. The aim is to create a large data set on which epidemic modelling, infection spread metrics and socio-economic impact of the disease.

The primary innovation in the solution is a diagnosis device that enables collection of a significant amount of data about prevalence of TB in India and other developing countries that has been missing from data banks around the world. Thus, number of TB samples diagnosed on our instrument will be an initial aspect of judging the impact. We evaluate the opportunities that our solutions bring in with the fact that potential modelling and data mining opens doors to meticulous planning by private and government agencies on tackling with the TB crisis. Our primary measure of success would be the dependance of the local government on the data we have generated to create any intervention policies. 

Our aim is to stop the spread of MDR-TB and thus measuring the prevalence of MDR-TB before and after deployment of the solution will also define the impact of our solution. Through our epidemiological data research, we will present potential metrics that can be used for policy making, such as symptom to detection time, infection mortality, % of undetected cases, average disease severity on patient arrival, average time for treatment for all districts. Success in implementation of this metrics will also in-turn be a successful application of our research.  

The solution proposed here is developed with members from different academic and industrial fields, which provides us with a wider perspective to deal with the crisis at hand. But to support such an initiative and provide a jump start, to transform this idea into reality, ample financial and technical resources are required. At this stage, we are primarily looking for financial support for expansion of our team which will help in industrial prototyping of the device as well as work on data modelling. The team until now has been very successful in reaching out for collaborations across govt and private sectors, yet it is of grave importance that we are able to garner support from the local public health agencies in-order to be able to collect data and conduct pilot projects.

University of Cambridge - Physics, Engineering, Criminology Departments

IIT Roorkee-Physics Department

Arete Advisors

Collaborator affiliations-

CSIR- Institute of Genetics-New Delhi

SGV University-Jaipur

AIIMS-Jodhpur

Lok Manya Tilak Hospital Mumbai and WHO TB Elimination Core Group

India Stack, Avanti Finance and Indihood

The foundation of our solution is a team that has been built on collaboration from different fields of optical engineering, epidemiology, data science and social sciences. Our primary aim was to build an end-to-end solution which will cover all aspects of the crisis against TB. Through the Trinity Challenge we are hoping to extend our collaborations and expand our team. We seek mentors who can provide feedback on our solutions. We also seek support in resources which includes monitory, technical, guidance, potential support from government agencies, impact investment firms and other intermediaries that help us achieve our goals in fighting TB with a revolutionary epidemic management research strategy. We inherently plan to develop an open-source research pathway through which our data and models can be utilized for different data science studies to understand explain these epidemics solutions further.

Bill and Melinda Gates Foundation- In developing countries, it focuses on improving people's health and giving them the chance to lift themselves out of hunger and extreme poverty. The collaborations and impact that the organization has developed over India in the last decades have been phenomenal. We would request collaborative work with the foundation and setting up our resources. We are specially looking forward for support in conducting a pilot project in 2 districts to prove and improve our solution. For this, we seek support in gathering resources and we believe that we can make noteworthy progress with help of such an organization.