AI Tuberculosis Screening System

In 2020, an estimated 9.9 million people fell ill with tuberculosis (TB), of which a staggering 4.1 million were left undiagnosed and thus untreated, resulting in 1.5 million lives lost to this curable disease. In fact in 2020 Tb deaths rose for the first time in a decade, due to reduced diagnosis resources during the Covid-19 pandemic. Much like most of the high TB burden countries, South Africa adopts a subjective symptom-based TB screening method that is only 77% sensitive, this means that 23% of TB positives are missed by the TB diagnostic algorithm from the get-go. The resulting delayed or absent treatment extends the patient’s suffering in both time and severity, significantly increase their chance of dying, and causes the patient to remain contagious for longer. In fact, 45% of untreated TB patients die from the disease, patients coinfected with HIV are almost certain to die if left untreated for TB. 80 to 90% of all TB cases are infections of the lungs, this most contagious classification is called pulmonary tuberculosis (PTB).

The TB epidemic also inflicts a significant financial burden on a county’s national health system. The same subjective symptom-based TB screening method, as currently used in South Africa, is only 65% specific. This means 35% of every TB Negative patient screened is sent for expensive bacteriological confirmed testing. In South Africa, that relates to roughly 90% of these expensive TB tests being deemed unnecessary. This costs the already resource-limited national health system both time and money. Much like some of the high TB burden countries, South Africa currently spends 20% of their entire TB budget on final bacteriological confirmation tests alone.

The current TB diagnostic algorithm is reliant on patients self-presenting at the clinic. Due to screening inefficiency, active TB case detection (detection outside the clinic structure) has proven to be too expensive and resource-intensive to warrant the few extra TB positives captured.

Our solution is an Artificial intelligence (AI) enabled, pulmonary tuberculosis (PTB) screening tool, which analyses digital stethoscope recordings to indicate whether a patient should be sent for further TB bacteriological confirmation testing or not.

Our solution consists of our custom build digital stethoscope and accompanying firmware, an app designed for tablets with an easy-to-follow interactive graphical screening guide, a locally executable AI engine that is run by the app, and cloud data storage. 

Our solution is a consumable-free (and disposal-free), portable, point-of-care (POC) screening tool, with no waiting period for results. Our solution is set to halve the number of TB-positive patients currently missed due to the currently employed TB screening tool (as outlined in the problem statement), while at the same time halving the number of TB negatives sent for resource-intensive secondary TB testing unnecessarily.

This 2min training video provides an overview of our solution - note this app version is set up for data collection with a laptop for one of our blind clinical studies, hence no TB prediction is provided to the user.

This screening method upgrade will result in an efficiency gain set to increase the screening sensitivity from 77% to 90% (screening accuracy in TB-positive patients). This will roughly halve the number of TB positives currently missed by the system. Earlier diagnosis for these patients will significantly reduce their suffering, it will increase their treatment outcomes while having the potential to reduce the length of their treatment regimes (essential to minimize treatment defaulting). Earlier diagnosis and thus early treatment will also reduce the disease's contagious period, reducing the chance of disease transmission to already vulnerable family members. 

This screening method upgrade will result in an efficiency gain set to increase the screening specificity from 65% to 80% (screening accuracy in TB-negative patients). This will roughly halve the number of unnecessary expensive bacteriological confirmation tests (currently 90% of tests are done on TB-negative patients). This will save nurses much need time while saving the South African TB budget roughly 19.3 million USD annually on direct secondary TB tests.

These screening efficiency gains are likely to make active screening (detection outside the clinic structure) viable, thus ensuring the TB diagnostic algorithm is less reliant on patients having to self-present at clinics, ensuring these currently missed patients are serviced. 

Active TB infection strongly correlates to one's socioeconomic circumstances. Unfortunately, being infected with TB perpetuates this situation. The relative household cost of TB is enormous, often removing the primary breadwinner from work for extended periods or permanently. Delayed diagnosis and treatment exasperate this issue resulting in longer patient recovery and reconditioning. Early detection provided by our solution will increase patient integration rate back into society, preserving both their livelihoods and the livelihoods of their loved ones.

I completed my Master's in Mech. Eng. After which I developed novel heart valve replacements and delivery systems that were designed primarily to serve patients suffering from rheumatic heart disease. 

Johan has a Master's in Biomaterials. After working in the medical product development field he switched careers and trained as a data scientist, specializing in AI. He has previously worked on multiple audio interpretation projects. 

Mark has a BSc in Electrical Eng. and has 33 years of experience launching and managing business ventures, the last of which is a specialized healthcare facility. 

As a technical co-foundering team we were able to develop our MVP quickly and cheaply and are well-equipped to manage the appropriate development partners during future product refinement. Collectively we have 14 years of medical device development experience, and I have 6 internationally granted medical patents. Johan has developed and commercially launched numerous AI engines. While Mark has hired 100s of nurses and health care workers providing essential insight into the end user.

We have and continue to be supported by phenomenal advisors, ranging from a national healthcare insider, disease specialist, regulatory advisor, successful multi-software startup founders, and VC specializing in hardware commercialization. 

We have completed a data collection study with over 30 000 validated lung sounds (50% TB positive 50% TB negative but TB suggestive). From this data set, we have trained and tested our proprietary AI models. We have received 3rd party technology support off the back of these study results from FINDorg, FEND-TB consortium and SAMRC. In partnership with FEND -TB we are conducting an international study in Vietnam, Peru, Uganda and South Africa to see how well our tech generalizes in other demographics.

Hardware wise we have designed and produced 2 digital stethoscope versions which have been used and tested on thousands of patients. We are in the process of finalizing and testing our Go to Market version.

Firmware we have developed and tested 2 versions, we developing a final version to accommodate a new battery monitoring feature in our new stethoscope.

We have developed multiple application versions throughout the clinical studies. The UX has matured via user feedback. Technical formalities such as ensuring remote app updates and integrating the AI engine into the app are the features still in development for the end of June. Cloud syncing, storage, and security have been developed and used throughout the data collection studies. 

The AI model has POC was conducted in 2022 March. we have since increased the training set 7 fold improved the data preprocessing protocols and increased our detection accuracy and established significant test numbers in sub-groups like HIV-positive, smokers or patients that previously had TB. 

Commercially we are set to start a pilot with a major mining group in July to routinely screen their employees for TB.

We have established a Quality Management System (QMS) and are making use of a digital system to ensure we remain compliant with ISO 13485.

From a regulatory perspective, we will be legally able to provide our screening services in South Africa in June this year when our SAHPRA license is expected to come through.

We are currently in the final product refinement stage. Unfortunately, we are not yet serving the target beneficiaries.

As part of our regulatory strategy to expand to other jurisdictions, we will be applying for FDA approval, likely through the 510K approval path. I believe being part of this program will expose us to the correct partners to guide and support us through our regulatory pathway. 

I believe this may expose us to MIT's ML/AI expertise which is currently rated in the top 5 academic institutions for these divisions globally.

We have been privileged to be supported by Rutgers Uni, Emory Uni and Nelson Mandela Uni throughout our development and witnessed the game-changing impact and weight of support from leading institutions. 

Innovation has been crucial but not the goal itself. Beyond detection accuracy, our product design process has been centered around usability and operating optimally in bustling rural clinic environments. It is these constraints that has given birth to the usage of existing technology in an innovative way.

Nursing and especially community health worker don’t typically know all the names of the auscultation positions. Therefore our apps Graphical User Interface (GUI), includes an easy-to-use touchscreen, which guides the health worker through the data collection process, indicating when and where to record. Furthermore, an image of an inflating and deflating balloon is used to guide the patient's breathing to ensure data consistency. 

To accommodate times of load shedding (routinely scheduled power outages) or active TB detection fieldwork, our AI engine is integrated into the app for offline usage.

Our cordless digital stethoscope is fitted with two microphones, one aimed at the auscultation diaphragm and the other pointed outwards to pick up ambient noise. The bustling clinic background noise is digitally subtracted from the auscultation signal prior to the AI analysis step to maximize accuracy.

Image recognition is one of the most researched and advanced sectors in AI. To leverage this established technology, we convert the auscultation recordings (sound files) into a spectrogram (a frequency domain visual representation of a sound file). We then use established image analysis neural network architecture to analyze these spectrogram “images”. 

Our TB data collection has been pivotal to developing our solution. To overcome challenging COVID-19 restrictions and funding constraints we have needed to innovate in this area as well. We developed a digital data collection system, with synced secure cloud-based data backups. Furthermore, we have developed an automatic participant booking system (in three native languages), where potential participants initiate contact by WhatsApping ‘hi’, they are then automatically checked for eligibility on the National Health Laboratory Service (NHLS) and booked an appointment on Google calendar.

Impact goals - next year:
1. Access to TB testing at 100 clinics equates to 57 000 patents per month.

Low LSM private clinic chains (Unjani clinics) seek to fill state clinic shortfalls. Currently, these institutions are unable to provide TB testing because the cost of lab testing on the customer profile is simply too exorbitant. Low LSM health insurers are unable to warrant adding TB lab testing to their coverage because the trigger for a test in an inefficient symptom-based TB questionnaire. Our screening tool bridges this gap, allowing these health insurers to cover TB lab tests if our screening tool indicates the patient is in fact at risk of having TB.

We are in conversation with both Unjani Clinics, Dis-Chem Health insures to make this a reality.

2. Capture an extra 10% of TB positives mine workers currently missed.

By enabling a cheaper more frequent screening option for the mining groups we are able to identify TB positives sooner, minimizing the spread of TB in the confined working environments. TB is currently 3 to 4 times more prevent in mine workers.

We are in the process of finalizing our first commercial pilot with South Africa's largest mining group. 

Impact goals - next 5 years:

1. Achieving an 80% TB detection rate across the whole of South Africa

This requires rolling our solution out in the national Department of Health clinics. how we will achieve this is outlined in the plan to become financially sustainable. Implemented throughout national clinics will ensure all state clinic patients are effectively and routinely screened for TB.

These screening efficiency gains also make active field screening (detection outside the clinic structure) viable, thus ensuring the TB diagnostic algorithm is less reliant on patients having to self-present at clinics, ensuring these currently missed patients are serviced.

2. Servicing at least one other high TB-burdened country.

Leveraging our South African results, we intend to extend our impact to other countries suffering under the TB pandemic.

Measuring Progress

All impact is based on the increase of both screening sensitivity and specificity and we therefore track these metrics through our AI model optimization and international clinical studies.

Next Year

1. Access to TB testing at 100 clinics equates to 57 000 patents per month.

- Track the number of private clinics fitted with our TB screening tool. 

2. Capture an extra 10% of TB positives mine workers currently missed.

- During our mining pilots using our TB screening solution we send employees who are identified as TB suggestive for lab tests. Our screening tool also uses the same symptom-based questions as part of the diagnostic algorithm. So after each month, we will report on the number of patients that we identified that were asymptomatic (symptom-free). These asymptomatic Tb-patients are currently missed by the traditional TB screening method. We are able to then clearly show what proportion of additional patients we were able to detect.

Next 5 years

1. Achieving an 80% TB detection rate across the whole of South Africa

This metric is provided and tracked by the WHO and StopTB partnership. It is defined as the number of TB-positive cases notified for treatment vs the estimated TB incident rate. Which for 2021 was only 57% (largely due to reduced TB testing because of Covid-19). This metric will be reviewed on a year-to-year basis.

Root Problem:

• Inaccurate TB screening methods.
  • Sensitivity > 77% -> 23% positives missed 
  • Specificity > 65% -> 35% Negatives sent for TB tests unnessearially
• Reliance on TB patients' self-parenting at the clinic. 
• The socio-economic living conditions

Actions (inputs):

Develop and rollout new TB screening method. Main technical requirements: 

• Sensitivity > 90%. 
• Specificity > 80%. 
•  Portable.

Link up with contact tracing and active screening initiatives.

Results (Outputs):

Increased Sensitivity -> Halve the number of TB positives currently missed -> Earlier TB detection.

Increases Specificity -> Halve the number of TB negatives sent for lab tests -> Reduce TB resource waste.

Portable (and accurate) -> Non-Clinic based active TB screening technically possible and financially viable -> Non-Clinic presenting TB positives identified earlier.

Outcomes/Long term goals:

Early detection -> earlier treatment -> better treatment outcomes -> Improved health. 

Non-Clinic presenting TB positives identified early. -> earlier treatment -> better treatment outcomes -> Improved health. 

Earlier treatment -> reduced contagious period -> reduced spread of the TB -> Improved community health. 

Reduce TB resource waste. -> Redirection of TB budget to medication adherence and other health necessities -> better treatment outcomes -> Improved health. 

Improved health -> Fewer costs & ability to work -> better financial outcomes.

Ethos:

Equity in Medical Care

Our proprietary AI model in our solution is the core technology. 

The AI model used is a residual neural network architecture. We use ResNet34 in particular. This architecture allows for the training of exceptionally deep neural networks, making it extremely well-suited for image classification tasks. We've leveraged this fundamental breakthrough by converting our lung sound files to image-like spectrograms using a Fourier transformation. We then trained this model on over 30 000 validated lung sounds or "images" (Converted audio recordings).

Diversity is essential to minimize biases and blindspots. We as the leadership group recognize the importance of this. It is our intention to ensure we developed a diverse and representative team as we grow.

Our business is centred around creating equity in medical care, we plan to mirror this ethos in our hiring policies.

Customer: National Department of Health - TB management programs. Occupational Health divisions of entities with large work forces (mines).

Beneficiaries: TB patients. employees (Factory & Mineworkers)

User: TB Nurses Community Health Workers.

Business model (Values based on the South African market only).

Per unit license fee revenue model. We maintain hardware ownership - facilitating national adoption due to no capital requirements. 

Customer Benefits: (based on TB detection accuracies achieved during the pilot study 80% and 83% sen and spec, respectively)

• 23K more TB positives captured P.A
• USD 19.3M saved on unnecessary secondary TB lab tests. 
• USD 10M net saving for the DoH, after the cost of service, has been subtracted.

User Benefits: 

• Halve the number of sputum sample collections from TB-negative patients req.
• Empowered, extra diagnostic capability.

Patient Benefits:

• Detected earlier, treated earlier, improved treatment outcomes.

Using funds raises through investment capital and grants we are finalizing out go-to-market product versions.

Use B2B contacts to prove ROI to National departments of health

We are proving our business case in the private occupational health sector (early adopting market). We are in the process of developing commercial pilot studies with a few key large entities, a large mining group, and a low LSM serving private clinic chain. We are pricing these at 1.5 cost price and doing so on a fix term contract basis. So far, because of the cost-saving impact our solution has, we've seen a great response, with our largest pilot prospect pushing to expand the pilot deal into an equity investment as well. If the commercial pilots are successful and convert prospects into clients we will revert to a license fee-based revenue model with the private clients as well.

Use state health economics studies to prove ROI to National departments of health

At the end of this year, we are conducting a health economics study in the state clinics to prove our financial and life-saving value to the relevant National departments of health decision-makers. 

Use TB activist groups and organizations to assist in lobbying for National TB management policy change.

There are strong national and international TB activist groups in South Africa, we are fostering relationships with these organizations and ensuring we co-develop the solutions to maximize support during the lobbying phase. 

Using WHO endorsement to drive National TB management policy change.

Our international study partners are direct advisors to the WHO on Tb management policies, we have joined forces to ensure we develop sufficient clinical evidence to be WHO-endorsed. South Africa closely mimics the WHO's policy decisions when it comes to TB management.

Although the Private sector revenue alone is sufficient to become financial sustainability, however, it will not be able to fully realize our impact goals. We, therefore, see the private market along with the other key market drivers as essential stepping stones to breaking into the public health sector, where our solution is most suited and accessible to truly fragile communities.

We then intend to charge the state a monthly per-unit license fee to use our product, the license fee of ~USD 200 is sufficient to maintain healthy operational cash flows and enable us to reinvest into other diagnostic capabilities.

A South African state contract will act as a gateway to other international service contracts. 

To fund development and clinical studies we have raised 260 000 USD in pre-seed investment.

We have won 26 000 USD in prize money.

We have been provided 52 000 USD in the form of a grant from the SAMRC.

We have partnered with FEND-TB who are covering the full cost of our international TB detection validation study. 

We have partnered with Rutgers University and the NIH's NIAID division to co-fund a new study in Uganda specifically in testing in asymptomatic TB patients.