AI-enabled Tuberculosis Screening and Prisoner Health

Tuberculosis (TB) is a communicable disease that is a major cause of ill health and one of the leading causes of death worldwide. Until the coronavirus (COVID-19) pandemic, TB was the leading cause of death from a single infectious agent, ranking above HIV/AIDS. A total of 1.6 million people died from TB in 2021, including 187 000 people with HIV (WHO, 2022). 

Globally, the estimated number of deaths from TB increased between 2019 and 2021, and an estimated 10.6 million people fell ill with TB in 2021, an increase of 4.5% from 10.1 million in 2020. The TB incidence rate rose by 3.6% between 2020 and 2021, reversing declines of about 2% per year for most of the previous 2 decades. The net reduction from 2015 to 2021 was 10%, only halfway to the first milestone of the End TB Strategy, and a step toward achieving Sustainable Development Goal (SDG) 3.3 aimed at ending TB by 2030 (WHO, 2022).  

 Low-resource countries are particularly vulnerable to TB.  The TB burden amongst incarcerated individuals has been rising exponentially, and particularly in Africa (by almost 29% from 2000-2018). TB has emerged as one of the major public health concerns across the prisons in Africa. For example, statistics showed that the TB case notification rate in Malawian prisons was 835 per 100,000 (the general population was 346/100,000). Increased rates of infection occur due to prisons being overcrowded, poorly ventilated, and under-resourced, with limited access to prompt TB diagnostics and treatment which makes them an ideal congregated setting for TB transmission (Cords et al., 2021; Dara et al., 2015; Kanyerere et al., 2012). Prisons in low-resource countries have higher TB notification rates as compared to the general population, ranging from 11 to 81 times higher.  

Prison settings increase TB transmission to family and staff members when engaging with infected detainees and present a public health challenge for the general population during the period of incarceration and upon release. To accelerate the integration of the TB positive individuals in the treatment care cascade of the National Programs; accurate, quick, and timely screening especially in the absence of specialist doctors (as prisons have a significant dearth of health workforce) is required to reduce the TB burden and facilitate early access to diagnosis and treatment. (Dara et al., n.d., 2015)   
 
 

Our proposed solution aims to address this gap by providing screening capabilities and reducing the turnaround time from screening to treatment access thereby contributing to the reduction of TB transmission and burden in these resource-constrained settings.   

Our solution in identifying and treating people infected with TB in prisons, and preventing disease spread is both easy to adopt and effective way. We use computer-assisted diagnostic technology to rapidly recognize over thirty lung health conditions. qXR is Qure.ai’s CE Class IIb AI-based Chest X-Ray processing tool. Our solution has been trained on more than 4 million Chest X-rays in diverse settings and geographies and deployed across 50+ countries and 600+ sites (many of which have little or no access to healthcare), and engaging key vulnerable populations representative of different demographics(e.g. prisons, nomadic population, people living with HIV, communities in rural and remote areas etc.). qXR aids in detecting comprehensive lung health findings on a chest X-ray in less than a minute. This reduces the chances of late diagnosis, under-diagnosis and potential misdiagnosis and helps ensure better patient healthcare delivery. qXR has been specifically developed to augment human expertise with the power of AI.  The qXR-TB program can identify multiple signs suggesting pulmonary TB on a chest X-ray(CXR), including opacity, enlarged or calcified lymph nodes, pleural effusion, blunted CP angle, or fibrosis. Based on the presence of these signs, the algorithm flags the CXR as TB presumptive. At sensitivity above 95% and 72% (69-75%) specificity qXR met the FIND’s Target Product Profile criteria for the WHO TB Triage test recommendation, supporting the evaluation studies which guided the WHO recommendation - “Among individuals aged 15 years and older in populations, computer-aided detection software programmes may be used in place of human readers for interpreting digital chest X-rays for screening and triage for TB disease.”(Qin et al., 2019) qXR can detect and localize multiple other abnormalities in the chest X-rays as well. It has been featured in 40+ peer-reviewed publications and has been used and recommended by multiple organisations across the globe including the UK NHS Trusts, the StopTB Partnership, FIND, and PATH. qXR is accompanied by qTrack – an end-to-end platform for TB screening, program, and case management. It tracks the end-to-end journey of a case in the TB Care Cascade to effectively monitor and establish linkage between various stages and narrow the gaps. 

Our solution qXR aims to serve the underserved incarcerated population in the high TB-burden prison settings of Africa. These prison settings are significantly under-resourced with limited access to timely diagnostics and access to treatment as well as a mobile population thereby resulting in rapid transmission and increased morbidity and mortality due to TB (Telisinghe et al., 2014). AI-based screening of chest x-rays will help in the triaging of the TB presumptive individuals as well as asymptomatic cases in less than a minute in the absence of the specialist which facilitates on-the-spot isolation and sputum collection for microbiological testing thereby reducing the turnaround time from screening to access to the treatment.  The identification of the people identified as TB presumptive also aids in reducing the unnecessary wastage of microbiological testing resources in congregated prison settings.     We are interested to accelerate our efforts by initiating our efforts in Malawi (complementary to our programs running in Central African Republic(CAR) and Haiti) where there are significant health needs for people in places of detention. In Malawi, there are 33 prisons with a population of over 18,000 individuals. The key beneficiaries of our public health-promoting interventions include the incarcerated population, their families, and the staff who work in and administrate the prison locations and the communities of Malawi. There are currently only 16 dedicated healthcare staff who provide care across the prison system in Malawi Prison Service.  These staffs do not have the capacity to meet the needs of the incarcerated population, access to the training and equipment required to test for many infectious diseases. Our solution involves using the qXR in tandem with a mobile x-ray to access all of the prison locations in Malawi and to screen the total prison population and staff group for TB. This technology and treatment intervention is not currently available to the detainees and staff, and symptom screening rarely occurs due to the low numbers of staff, and the absence of equipment and training. Additional stakeholders that our solution serve is:   
1. Healthcare workers in the prison: The shortening of the treatment pathways will result in the utilisation of the limited resources found in the prison settings thereby empowering the healthcare workers present in the prisons and giving them the tools to ensure equitable distribution of testing and pharmacological resources. 

2. Prison authorities – This solution empowers the prison authorities in ensuring the health and well-being of staff and detainees in prison settings and that the health of both of these groups is improved. This will reduce preventable deaths in custody and ensure that the high rates of Tuberculosis prevalent in these settings will reduce by ensuring that the limited resources available to prison authorities and their treatment linkages are utilised judiciously. 

3. Larger community – When prisoners are released into the community without testing for TB the disease spread continues unchecked. Testing staff and prisoners for TB while within the prison environment provides an ideal opportunity to test, isolate, treat, and prevent further transmission of the disease.

Qure.ai offers AI-powered medical image reading solutions for X-rays and other modalities and works with a wide range of partners to deploy the technology and achieve maximum impact for key populations. Our preferred partner for deployment in places of detention is Health through Walls. Our solution is deployed across 50+ countries,600+ sites, and we impact 3.5 million lives yearly through our solutions by engaging with communities and vulnerable populations across the globe. At Qure.ai, our mission is to make healthcare more accessible and affordable using the power of deep learning  To realise this vision, we have had experience serving in low and middle-income countries and various fragile populations across Asia and Africa including nomadic pastoralists in Nigeria, slum dwellers during the COVID-19 pandemic in Mumbai, India, and prisoners in violence-affected Haiti with our partner, Health through Walls (HtW) bringing in their implementation and operational expertise for the prison context.  We are keen to replicate this model in Malawi and other low-resource countries that would benefit from our partnership.  HtW has nearly 20 years of experience in establishing, managing, and evaluating prison healthcare activities in developing countries. HtW is committed to providing technical assistance and training in prison healthcare worldwide and helping to build locally “owned,” planned, and implemented sustainable initiatives.The mission is to assist low-income countries in implementing sustainable improvements in the healthcare services of their prisons, with a primary focus on the identification, prevention and management of TB, HIV and other infectious diseases. Qure.ai and HtW currently partner together in Mozambique, the Central African Republic, and Haiti, and are currently exploring opportunities to assist the prison authorities in Malawi, the Democratic Republic of the Congo as well as Brazil, Guyana, and Barbados gradually. 
In 2022, Qure.ai and Health through Walls undertook a pilot project of our proposed innovation to enable rapid TB triage in prisons using Artificial Intelligence to measure the impact of our work. Mass TB screenings and active case findings were conducted in overcrowded prisons of Haiti with mobile digital X-ray units, which were equipped with qXR. Persons identified with chest radiographs suggestive of TB by qXR were immediately isolated from others to reduce the spread and taken forward for confirmatory testing and treatment initiation. The disease management software qTrack aided in maintaining electronic medical records and tracks patient journeys until treatment completion. We also conducted an analysis of 166 participants of which 62.6% were found to be TB presumptive by qXR, wherein 77.9% were microbiologically confirmed. The time for radiologist review of CXR was one week, while the qXR results were instantly available.

AI-based Chest X-ray Tuberculosis screening consisting of different settings, populations and contexts: 
Qure.ai has partnered with various stakeholders globally to accelerate the screening and diagnosis of Tuberculosis and has deployed the AI solution across 440 sites screening about > 2 million scans to date.  
 
AI-based Chest x-ray TB screening in prison settings in partnership with HTW:  
Locations- CAR, Haiti 
The number of scans processed to date in CAR is more than 1000 while the number of individuals screened across 5 prisons in Haiti is more than 5000.

Qure.ai’s mission is to use AI to enable affordable, accessible and timely care across the globe. Our mission has fuelled our efforts in striving to reach vulnerable populations in varied settings and geographies to address the TB burden. Our partnership with the Health Through Walls organisation with extensive experience in facilitating healthcare in prison settings has been crucial in integrating technology into the prison’s healthcare system in Haiti and expanding in Africa in strengthening TB screening and diagnosis interventions. Our organisations together envision to scale-up our solution in the underserved and resource-constrained high TB burden settings across Africa and other regions gradually. In partnership with Health Through Walls, we have started to strengthen our foothold in the prison of  CAR and Mozambique. Through stakeholder engagements, Malawi has also been identified with a high TB burden prison setting. Prisons in Malawi had a high TB annual case notification rate (835/100,000) which is significantly higher than the general population (346/100,000). These prisons also have low microbiological testing resources wherein most tests need to be referred to nearby facilities or district hospitals increasing the turnaround time for testing and initiation of treatment. (Kanyerere et al., 2012) 

From our learning and experiences with the implementation in the prison of Haiti, we look forward to the mentorship from the challenge team to scale up our solution in this geography with the specific mentorship focus on creating consortiums/partnerships, impact measurement and evaluation and access to investors/funders for the long term sustainability. We plan to implement a pilot project to understand the facilitators and barriers in this region and demonstrate the proof-of-concept to the prison authorities for their further support in integrating our solution for the TB screening of the inmates across all the prisons. Mentorship and guidance from the experts in the challenge team would aid us in implementing and planning the scale-up of our intervention in Malawi, CAR, Mozambique and largely in the challenging African geography. Additionally, scaling up our intervention in the African geography would require consolidated efforts of a stronger consortium ensuring end-to-end TB management support which we aim to create with challenge team’s support by accessing their network of organisations and key experts. The impact measurement of our interventions with support from challenge team would help us strengthen our effort and facilitate access to funders and investors for the scale-up, helping us reach this vulnerable population and reducing the burden of TB.

Suspected symptom-based screening has been a key component for active case finding for TB. But its sensitivity and specificity have been low wherein asymptomatic cases could also be missed. To improve the accuracy of the active case finding WHO recommended usage of CXRs for screening of TB. However, the utility of Chest X-Rays to diagnose TB-affected people is limited due to the unavailability of radiologists or skilled medical professionals in low-resource settings. In this scenario using an AI-based screening tool which identifies potential abnormalities on a Chest X-Ray, could be an optimal way to screen for TB in under-resourced settings addressing the human resource shortage. (Cao et al., 2021) AI-based chest X-ray screening improves the accuracy of identification of TB findings on X-rays even in asymptomatic patients. The turnaround time for chest X-ray screening is reduced to less than a minute, thereby accelerating the consequent processes in the TB care cascade.  

AI as a screening, triage and clinical decision-making tool includes the following benefits: 

• Reduction in reporting time for reading CXRs  
• Reduction in confirmatory microbiological test consumption  
• Reduction in Turn Around Time for Diagnosis and Treatment  
• Increase in the number of Additional TB cases Detected  
• Increase in the identification of Asymptomatic Cases  
• Increase in cases identified due to clinical diagnosis  
• Reduction in overall cost for screening and diagnosis  

These benefits are layered across the spectrum of the TB care cascade and its impact creates a ripple effect on ultimately strengthening the health system and its building blocks (service delivery, health workforce, technology).(Health System Building Blocks, n.d.) 
In addition to the technical expertise, Qure.ai’s partnership with Health through Walls in delivering our solution to the fragile context of the prison settings with their implementation and operational expertise strengthens our solution and approach through a better understanding of the local contexts and stakeholder engagement thus integrating a bottoms-up approach in reducing the TB burden.

Impact in the next year: 

 
Qure.ai and Health through Walls plan on expanding on their experience of delivering AI-based TB diagnosis solutions, especially focusing on the prison population. This will in part be executed by developing a model of a successful TB diagnosis and notification program which shall be buttressed with robust impact evaluation to clearly quantify the impact of such a solution being introduced in fragile populations. This becomes especially important to bring about an improvement at the health system level, with technology demonstrated to bring the maximum impact and reduce the Turnaround Time and consequently, the investment required to increase TB case notification.  
 

Next five years: 

1. Scale up of the solution in high TB burden prison settings in low-resource prison settings of Africa, Brazil, Guyana, and Barbados. 
2. In addition to Tuberculosis, screening of other lung health abnormalities on chest x-ray through qXR; implementation of an Integrated Lung Health Screening program with end-to-end screening and treatment linkages in the prison settings. 
3. Identification of other vulnerable/ fragile populations with high TB burden and low notification rates in LMICs and stakeholder/partner engagement in fragile contexts (such as refugees, IDPs, and miners,) to scale up the implementation of an AI-based chest x-ray screening program for tuberculosis and other lung abnormalities. 

1. Number of prisons qXR-TB is deployed at.  
2. Number of Health professionals trained on qXR-TB  
3. Number of CXRs screened through qXR-TB  
4. Number of CXRs flagged as positive for TB  
5. Number of all CXRs flagged as negative for TB  
6. Mean Turnaround time from X-Ray generation to Sputum Test  
7. Number of qXR-TB presumptive scans TB positive after sputum test  
8. Number of TB patients screened annually   
9. Number of TB-positive patients identified.  
10. Mean overall cost to diagnose TB patient. 
11.  Number of patients initiating TPT 
12. The number of patients successfully integrated into the treatment cascade. 
13. The number of patients who successfully completed the treatment regime. 
14. Experience (challenges and facilitators) of frontline workers and healthcare professionals with our AI solution. 

References -

•Cords, O., Martinez, L., Warren, J. L., O’Marr, J. M., Walter, K. S., Cohen, T., Zheng, J., Ko, A. I., Croda, J., & Andrews, J. R. (2021). Incidence and prevalence of tuberculosis in incarcerated populations: a systematic review and meta-analysis. The Lancet. Public Health, 6(5), e300. https://doi.org/10.1016/S2468-...

•Dara, M., Acosta, C. D., Melchers, N. V. S. V., Al-Darraji, H. A. A., Chorgoliani, D., Reyes, H., Centis, R., Sotgiu, G., D’Ambrosio, L., Chadha, S. S., & Migliori, G. B. (2015). Tuberculosis control in prisons: current situation and research gaps. International Journal of Infectious Diseases, 32, 111–117. https://doi.org/10.1016/J.IJID...

•Dara, M., Chorgoliani, D., & De Colombani, P. (n.d.). 8. TB prevention and control care in prisons Key points.

•Kanyerere, H. S., Banda, R. P., Gausi, F., Salaniponi, F. M., Harries, A. D., Mpunga, J., Banda, H. M., Munthali, C., & Ndindi, H. (2012a). Surveillance of tuberculosis in Malawian prisons. Public Health Action, 2(1), 10. https://doi.org/10.5588/PHA.11...

•Kanyerere, H. S., Banda, R. P., Gausi, F., Salaniponi, F. M., Harries, A. D., Mpunga, J., Banda, H. M., Munthali, C., & Ndindi, H. (2012b). Surveillance of tuberculosis in Malawian prisons. Public Health Action, 2(1), 10. https://doi.org/10.5588/PHA.11...

•Key Populations Brief: Prisoners | Stop TB Partnership. (n.d.). Retrieved April 19, 2023, from https://www.stoptb.org/key-pop...

•Khan, F. A., Majidulla, A., Tavaziva, G., Nazish, A., Abidi, S. K., Benedetti, A., Menzies, D., Johnston, J. C., Khan, A. J., & Saeed, S. (2020). Chest x-ray analysis with deep learning-based software as a triage test for pulmonary tuberculosis: a prospective study of diagnostic accuracy for culture-confirmed disease. The Lancet Digital Health, 2(11), e573–e581. https://doi.org/10.1016/S2589-...

•Soares, T. R., Dias De Oliveira, R., Liu, Y. E., Da, A., Santos, S., Pereira, P. C., Luma, S., Soares, R., Lissandra, M., De Oliveira, M., Park, C. M., Hwang, E. J., Andrews, J. R., & Croda, J. (2021). Evaluation of chest X-Ray with automated interpretation algorithms for mass tuberculosis screening in prisons. MedRxiv, 2021.12.29.21268238. https://doi.org/10.1101/2021.1...

•Telisinghe, L., Fielding, K. L., Malden, J. L., Hanifa, Y., Churchyard, G. J., Grant, A. D., & Charalambous, S. (2014). High Tuberculosis Prevalence in a South African Prison: The Need for Routine Tuberculosis Screening. PLoS ONE, 9(1). https://doi.org/10.1371/JOURNA...

• Velen, K., & Charalambous, S. (2021). Tuberculosis in prisons: an unintended sentence? The Lancet Public Health, 6(5), e263–e264. https://doi.org/10.1016/S2468-...

qXR is a radiology computer-aided analysis software intended for use as an aid during chest X-ray interpretation. The device uses artificial intelligence algorithms to analyze chest X-ray images in parallel to the ongoing standard of care image interpretation and provides a list of suspected findings on the chest X-ray. The solution trains and validates the system with large radiology data to automatically analyze and highlight suspected abnormal regions. The software application of the product works independently on cloud servers (as well as on-premises) and does not interfere with radiology IT workflow. The device can identify and localize the suspected abnormalities on chest X-rays. The user is presented with preview images highlighting the abnormal findings, that are meant for informational purposes only and not intended for diagnostic use. After an automated and non-interactive processing of the x-ray images, qXR software application generates a report and bounding boxes, if relevant around the region with abnormalities suspected. The device does not alter the original medical image. It is not intended to be used as a diagnostic device, a source of medical advice or to aid in determining patient management plans. The results of the device are intended to be used in conjunction with other patient information and based on professional judgment, to assist with the efficiency and accuracy of radiological image interpretation. Clinicians are responsible for viewing the original chest X-rays as per the standard of care.

Qure.ai treats all individuals fairly and impartially, without prejudice, and does not tolerate harassment in any form. We are committed to developing a working culture that is fair and 'inclusive' -enabling all employees to make their distinctive contributions to the benefit of the business. We are also determined to ensure that we extend this same openness to our suppliers, business partners and all our customers. We expect our managers to exercise leadership in this field by discouraging prejudice and by role-modelling appropriate behaviour. Qure.ai is an “equal opportunity” organization that prohibits discrimination or harassment based on race, colour, religion, national origin, sex, age, sexual orientation, marital status, citizenship status, or disability. 

Qure.ai's versatility and ability to quickly integrate into any client user’s existing software design or workflow, our technology/solutions can be applied to virtually any healthcare setting. Our clientele includes major metropolitan hospitals, managed care medical centers, and government departments/ministries of health. Since Qure.ai’s deep learning capabilities dramatically improve the time-to-diagnosis and aid in prioritizing cases, the technology is also well-suited to support small radiology departments in community hospitals and to assist teleradiology entities that typically manage thousands of x-ray images and scans (requiring prompt turnaround). 

Our products are constantly evolving to meet the needs of our partners reaching- vulnerable populations in hard-to-reach under-resourced- geographies and strengthening global commitments to improve health outcomes. A diverse spectrum of credible partners, a public health portfolio adept at meeting changing needs, strong deployment history/use cases, diligent customer success teams and evidence-based actions have ensured the organisation is on course with their mission. 

Qure.ai utilizes a pay-per-use model allowing low-volume clinics and radiology centers in suburban and rural areas to use the technology. The company is also equipped to execute on-premises deployment. In addition, the solutions are integrated with more than 10 of the most common radiology viewing platforms, which ensures seamless integration with existing software and workflow tools already in use by Qure.ai clients.

Qure.ai has achieved financial sustainability since its inception through multiple sources:  
1. Investment: Series C funding has been secured by Qure.ai through equity-based investment.   
2. Commercial partnerships: Qure.ai’s products have been procured globally for different disease areas such as from LTE, AstraZeneca, Fujifilm etc.  
3. Grants: Qure.ai has won several grants in the past including from SBRI, India Health Fund, NHS  and STOP TB.