CerviScanner

Cervical cancer starts in the cervix cells at the lower end of the uterus. This cancer is a pressing public health issue that disproportionately affects women in low-income African communities. In 2018, all but one of the top 20 countries worldwide with the highest burden of cervical cancer were in Africa. The empirical reality underscores that, although this type of cancer is among the most preventable, limited access to prevention, screening, and treatment services results in 90% of deaths. 

One of the most significant challenges in addressing this issue is the limited access to essential preventive healthcare services, including human papillomavirus (HPV) vaccinations and preventive screenings. In many African countries, healthcare infrastructure is either underdeveloped or inaccessible to impoverished people. This lack of access leads to late diagnoses, where treatment options are limited and less effective. The shortage of pathologists in Africa exacerbates a dire healthcare situation, particularly in cancer diagnosis and treatment. For instance, a report on radiation services in Nigeria—a country with a population of 160 million and an estimated 100,000 new cancer cases annually—revealed a staggering lack of specialized medical professionals. The entire nation had only 18 radiation oncologists, 8 medical physicists, and 18 radiation therapists to meet its radiation therapy needs . This scarcity often leads to misdiagnosis, as generalists who lack specialized training in pathology are left to handle complex cases. The result is delayed or incorrect treatment and a missed opportunity for early intervention. 

Early prevention, including HPV vaccination and early diagnosis, is crucial for cervical cancer. However, African countries are currently not meeting the World Health Organization (WHO)'s 2030 targets for prevention and treatment, including a 90% HPV vaccination rate by age 15, a 70% screening rate between ages 35 and 45, and a 90% treatment rate. The 5-year survival rate for cervical cancer in Africa is below 50% , which is in stark contrast to the survival rate of over 65% in developed countries. Innovative solutions are urgently needed to address unique challenges in low-income African communities. A multidimensional approach prioritizing early-phase screening for African women can significantly reduce mortality rates, aligning with WHO's ambitious goals.

We designed a mobile application, “CerviScanner”, to screen for cervical cancer among African women aged 21 to 69. The app works as an intelligent screening assistant for healthcare professionals (HPs). A HP places the smartphone camera lens over the eyepiece of a microscope, which is focused on a Pap smear slide. The app captures the image and uses a sophisticated Artificial Intelligence (AI) algorithm to analyze the cellular structures for abnormalities indicative of cervical cancer. The results are then displayed on the app's interface, providing immediate, actionable insights for the HP. The mobile app is a screening tool that eliminates the need for additional hardware or computer systems. We trained our app on thousands of cervical smear test images after a thorough four-phase research methodology. The latter included dataset collection, pre-processing, predictive model selection, and code implementation. Our algorithms have reached an accuracy rate of 89% in uncontrolled settings. 

Our demo is accessible here. 

Our initial target is 5 million women aged 21 to 69 in Tunisia. The next phase will expand the application to Kenya, targeting 16.8 million women at risk of developing cervical cancer , followed by an outreach to 4.35 million women in Rwanda. Ultimately, our objective is to focus on women in the remaining top ten Sub-Saharan African countries with the highest rates of cervical cancer, including Eswatini, Malawi, Zambia, Tanzania, Zimbabwe, Lesotho, Uganda, Comoros, Mozambique, and Guinea. 

A poignant example of the pressing need for our solution comes from South Africa, where women are typically asked to return to their primary healthcare clinic within 6 to 8 weeks just to receive Pap smear results. Those identified with abnormalities face an additional wait of 1 to 8 months for a colposcopy appointment. It is in this gap of time where the danger lies and where our solution intervenes. By implementing CerviScanner, the time from test to results can be reduced to 1-2 hours, allowing HPs to make immediate and informed decisions about the next steps in patient care. Our approach not only shortens the diagnosis timeline but also increases the volume of screenings, providing a significant boost in early detection rates and the potential to save lives.

As we look to bring preventive measures that are both cost-effective and sustainable, we anticipate a broad-scale impact that aligns with global health advancement goals. Our comprehensive approach is designed to be a game-changer in the regions we aim to serve. Over the next three years alone, we estimate reaching 2 million women across Africa, substantially improving access to care and contributing to the global fight against cervical cancer.

Medntech.org is distinctively equipped to revolutionize healthcare in Africa, with a foundation built upon firsthand experience and direct knowledge of the region’s culture and medical landscape. Our founder, who completed medical training in Tunisian hospitals and holds an MSc in Public Health, has an intimate understanding of the specific challenges and needs within the local healthcare system. Our other team members—currently practicing as a gynecologist and family doctor within the Tunisian healthcare system—give us the insight necessary to design and implement ground-breaking solutions, tailored to the needs of the communities we serve.

As a nonprofit organization driven by a public-health vision, our multidisciplinary team of local and international medical doctors, AI specialists, and public health experts brings diverse yet complementary skills to the table. Our commitment to health equity is reflected in our structure and approach, ensuring that our innovations serve not just as medical solutions but as tools to reduce healthcare disparities across our continent.

The Ministry of Health in Tunisia is more than an endorser, providing key resources and public health expertise to optimize the implementation process. Their involvement ensures our innovation aligns with national healthcare pathways, enhancing effectiveness and streamlining scale-up. This active collaboration is pivotal for rolling out our solution effectively.

These facets of our organization, including the practical medical expertise, the technological innovation, and the governmental collaboration, fortify our mission. They allow us to effectively bridge the gap between cutting-edge technology and the day-to-day realities of healthcare in Africa, enhancing accessibility and quality of care.

Our pilot phase was conducted in October 2023 in two prominent Tunisian hospitals, Hôpital Charles Nicolle and Hôpital La Rabta, targeting the detection of cervical cancer through our app. We focused on non-pregnant, sexually active women aged 21 to 69, excluding those with prior cervical procedures. In total, 189 women provided viable samples for analysis.

The hospitals were chosen for their high patient volumes, offering a diverse set of data for robust testing. With a consent process in place, we ensured participants were informed and willing. The sample size was based on the expected prevalence of cytological abnormalities, calculated to ensure statistical reliability for the app’s sensitivity and specificity.

Our app demonstrated an 89% accuracy rate, with a sensitivity of 90% and specificity of 89.33%, as corroborated by traditional medical examinations. This pilot phase has served to validate our app's effectiveness in a real-world clinical setting, marking a significant step toward improving early detection rates in underserved populations.

Our aspiration to revolutionize cervical cancer screening in Africa with CerviScanner app is fueled by a commitment to achieving 99% diagnostic accuracy using AI technology. This goal is inspired by MIT’s groundbreaking AI research in health, such as the Tyche model, which intelligently captures uncertainties in medical imaging, and initiatives like MIT’s work on automating feature identification in large datasets for improved medical decision-making. These efforts resonate with our vision to use AI to enhance patient outcomes, a pursuit we aim to advance through rigorous AI development and mentorship available via the Solve community. Additionally, our project is particularly aligned with the MIT Sloan Health Systems Initiative, which has developed web-based applications for clinical decision-making across various health domains, including prostate cancer. This parallel drives our focus on cervical cancer, underlining the importance of tailored digital health solutions in addressing specific medical challenges.

Furthermore, our team’s strong research background and interest in leveraging Solve’s resources for mentorship and partnerships position us to conduct impactful research and launch joint initiatives. Inspired by MIT’s comprehensive analysis of AI’s potential in medical diagnostics, we plan to contribute our insights and findings to both the academic and global health communities, enhancing the collective understanding and application of AI in healthcare.

Initiatives like the "Digital Health Challenge," which mobilizes companies with innovative solutions to address women health, are of particular interest to us. They exemplify the opportunities to connect with other solvers and engage with over 400 tech-focused social innovators. This network offers a critical opportunity for collaboration and partnership, essential for scaling our pilot from Tunisia across Africa. These alliances will help propel our technology to new heights, setting groundbreaking standards in healthcare delivery. As a venture led by a female founder, we are poised to contribute to and greatly benefit from this focus, leveraging cohort learning and expert guidance to close the gender equity gap in health.

While funding is indeed beneficial, the true value of joining Solve lies in the access it provides to leading minds in technology and social impact, including organizations like The Bill & Melinda Gates Foundation, whose commitments to global health mirror our own. This engagement will empower us to refine our technology and expand our impact, ultimately transforming the landscape of women's health across the African continent.

Our solution, CerviScanner, revolutionizes cervical cancer detection by leveraging AI technology to transform any smartphone into an effective screening tool. This innovative approach stands in stark contrast to costly alternatives such as telepathology, which can incur expenses up to $120M for software alone in a country like Kenya, and digital microscopes, averaging around $1,500 per unit—costs that are prohibitive for many healthcare sites within the African context.

In contrast, our app is accessible, offered for free, and integrates seamlessly with existing microscopes at no additional cost. This ensures widespread adoption is feasible and realistic across diverse healthcare settings. By utilizing smartphones, our app bypasses the need for expensive digitalizing equipment or new hardware. This not only reduces direct costs significantly compared to other methods but also speeds up the diagnosis process, eliminating the need for specialist intervention and the associated wait times.

Our app is designed to be compatible with a minimum camera resolution of 8 megapixels and works with both Android and iOS systems, ensuring broad usability. It functions effectively with conventional microscopes available in all hospital laboratories in Tunisia. CerviScanner is designed to function even when there is no internet connection. In such scenarios, the app will automatically save the microscope picture and store it locally on the device. Once an internet connection is re-established, the app will automatically upload the saved picture to the server. 

In terms of catalyzing broader positive impacts, our reach and data collection are exponential. Starting with 20 facilities in Tunisia and expanding to 80, with every facility conducting around 20 screenings/ day, we expect to conduct over 1 million tests across Africa within three years. This volume of data will refine our AI for greater precision, create cost savings for healthcare systems, and most importantly, save lives. The ripple effect of our app has the potential to set a new paradigm in healthcare efficiency and community empowerment, showcasing a sustainable model for global health initiatives.

CerviScanner leverages AI technology to enhance early detection of cervical cancer efficiently and affordably. Below is an outline of our Theory of change, detailed in the accompanying image, which describes how we plan to transform cervical cancer screening practices across Africa.

• Inputs: Our initiative leverages critical resources including funding, machine learning expertise, local healthcare partnerships, cloud infrastructure, and dedicated personnel.

• Activities: Key activities involve:
  • Refining our app’s accuracy.
  • Integrating user feedback for continuous improvement.
  • Collaborating with healthcare facilities to ensure seamless integration.
  • Forming an Expert Committee with diverse healthcare stakeholders.
  • Evaluating our impact using specific KPIs.

• Outputs: Initially, the app will be deployed in 20 facilities in Tunisia, expanding to 40 in Tunisia and Kenya by the second year, with a marked increase in screenings.

• Mid-term and ultimate outcomes: We aim to increase screening rates from their current level <30% to 50% by year one and 70% by year two. This objective resonates with the WHO’s recommendations. Ultimately, we strive to raise the five-year survival rate for cervical cancer from below 50% to at least 60%, bringing it closer to international standards. This will significantly reduce morbidity and mortality, and improve equitable access to cancer care across diverse communities.

Our solution’s impact goals are tightly interwoven with the UN Sustainable Development Goals, (SDG) and particularly SDG 3, underscoring our commitment to improving health outcomes through increased access to and quality of healthcare services. We will use these KPIs To measure our progress:

• Screening frequency: we monitor screenings through our app's backend, collecting real-time usage data to assess our reach and gauge the potential reduction in premature mortality.

• Diagnostic accuracy: to support universal health coverage as outlined in SDG target 3.8, diagnostic outcomes from our app are cross-referenced with medical records from healthcare facilities, ensuring accurate and reliable diagnostics.

• Early interventions: tracking early interventions in line with SDG target 3.4, we gather data from healthcare providers' records, enabling us to evaluate the effectiveness of early treatments that our app facilitates.

• Mortality rate: our contribution to SDG target 3.4 extends to studying mortality trends through public health records, which informs the long-term impact of our solution on reducing cervical cancer mortality rates.

• User satisfaction: we conduct periodic surveys with users and HPs to ensure the app delivers a satisfactory and effective user experience.

Data from these various sources allows us to closely monitor our KPIs, not only to measure our app’s direct impact but also to continually refine our approach. This ensures that as we scale, our strategies remain data-driven, user-focused, and aligned with the global and sustainable health advancements.

Our model has been trained using 2 robust publicly available datasets that include 1,415 images, categorized under The Bethesda System. These images, sourced from real-world clinical settings, ensure our model is both relevant and effectively tuned for precision in medical diagnostics.

The AI model operates through a series of sophisticated steps beginning with image preprocessing, where images are resized, cropped, and formatted to optimize analysis. Feature extraction follows, with the AI identifying crucial image characteristics such as edges, colors, and textures. These features are essential for the model to recognize various stages of cervical lesions. During the training phase, the model adjusts its parameters to reduce errors and improve its ability to predict pre-cancerous and cancerous conditions accurately.

Once trained, the model applies its refined algorithms to new images, making predictions based on learned patterns. Post-processing of predictions enhances accuracy and reliability by filtering out uncertain results and refining the diagnosis. Beyond simple image classification, the model also supports object detection, image segmentation, and even image restoration—capabilities that broaden its applicability and effectiveness in medical imaging tasks.

Supporting the AI model is a robust backend infrastructure designed around a microservices architecture orchestrated by Kubernetes, which ensures scalability and reliability. This setup manages various functions from handling user requests to processing complex data efficiently. Our system employs cloud services including Kubernetes for automated deployment and scaling, PostgreSQL for reliable data storage and management, and GPU-accelerated virtual machines that provide the necessary computational power for intensive data processing and model training.

To further enhance the algorithm, we have incorporated pilot data collected from 189 samples across two hospitals in Tunisia, enabling us to refine the model with direct input from local real-world scenarios. This practical validation helps ensure that our application is not only theoretically sound but also effective in everyday clinical use.

Our team dedicated to this solution consists of:

• Full-time staff:

  • 1 MD & Public health expert: Overseeing project alignment with objectives and providing research expertise.
  • 1 Intern and 2 volunteers: Focused on outreach, partnership establishment, and data analysis, with numbers varying based on project phase and current needs.
  • 1 Data scientist: Managing the technical development of the application.

• Part-time staff: Local project coordinators in Tunisia, 2 medical doctors who adapt and implement the project locally.

All team members are originally from Africa, providing an authentic perspective on the needs of the target regions.

Our dedicated team has been actively developing and refining CerviScanner for one year. Throughout 2023, we operated on a volunteer basis, channeling our collective expertise and passion into a successful pilot test of the app. Our commitment during this foundational year laid the groundwork for our app’s capabilities and set the stage for the subsequent scaling of our initiative.

Our team at Medntech embodies diversity in professional expertise and life experiences. All team members are originally from Africa, offering a genuine understanding of the region's unique healthcare challenges. We have diverse backgrounds coming from different walks of life  such as medicine, AI, and public health. We are steadfast in our commitment to furthering gender balance, especially as our work directly impacts women's health.

We prioritize inclusivity, ensuring that all team members have a voice. Our meetings are spaces where ideas can be expressed freely and without judgment. This is vital as it reflects our roots in and dedication to the LMICs we serve, ensuring that our implementation strategies are not only effective but also culturally sensitive.

Our actions to foster a diverse and equitable workplace include:

• Expanding our recruitment drive to increase female representation, recognizing the importance of women in our mission and in leadership roles (currently we have a gender ratio of 1:1) 

• Conducting team-building activities that reinforce our commitment to diverse viewpoints and collective unity.

• Collaborating closely with local communities to align our health solutions with cultural expectations and needs.

Medntech is a nonprofit organization that provides a free AI-powered cervical cancer screening app to healthcare providers in low-resource African countries. Our mission is to make critical healthcare services accessible to all, especially in regions where affordability and access to advanced technology are major challenges.We firmly believe that basic healthcare services, especially critical screenings like those for cervical cancer, should not be commercialized but should be accessible to all, particularly in regions where financial resources are major barriers. Thus, our mission is to enhance preventive healthcare, especially in regions where access to advanced technology are major challenges.

CerviScanner integrates seamlessly with existing medical infrastructure, requiring no additional equipment and working effectively even in areas with sporadic internet connectivity. This ensures that healthcare providers, including hospitals and clinics serving large populations at risk for cervical cancer, can improve diagnostic accuracy and patient outcomes without incurring extra costs.

To sustain our nonprofit operations and expand our impact, we are exploring a Software as a Service (SaaS) model in economically developed countries. This approach allows institutions with the budget for innovative health technologies to enhance their diagnostic tools while subsidizing the cost for users in lower-income regions. The revenue generated from this model supports our mission of providing the app for free where it is most needed and enables continuous improvement and expansion of the app’s capabilities.

Our innovation is transitioning from a volunteer-driven model to a hybrid revenue structure, incorporating public funding, strategic partnerships, and a commercial SaaS model to ensure financial sustainability while staying true to our mission of providing high-quality healthcare technology for free in low-resource settings. Currently, we are in the process of securing funds from USAID. As we move forward, we will actively pursue grants from major global health entities such as Global Affairs Canada, and the Bill and Melinda Gates Foundation, which are essential for funding our initial rollout, research activities, and pilot program expansions. We have also prepared a list of 20 potential funders to support our development and scaling efforts

Simultaneously, we are forging strategic partnerships to enhance our reach and impact. Working with organizations like The Canada-Africa Community Health Alliance (CACHA) and the Africa Cancer Foundation (ACF) provides us with logistical support and community engagement, as well as crucial insights into cancer care in Africa. These partnerships not only help secure additional funding but also strengthen our operational capabilities in our target communities.

While we are developing a SaaS model targeted at healthcare providers in developed countries to generate a sustainable revenue stream, this revenue will support the free distribution of our app in underserved areas. This approach aligns our commercial activities with our social mission and is designed to be cost-effective for users in developed markets, providing critical support for our free services elsewhere. Additionally, we plan to rely on donations and government support to further sustain our initiatives and expand our impact