HomeGermany-StartupsChallenging the status quo of healthcare: Jonas Muff tells us about the...

Challenging the status quo of healthcare: Jonas Muff tells us about the vision behind Vara

Founded in Germany in 2018, Vara is fuelled by the conviction that every woman around the world should have access to breast cancer screening. The AI-powered platform, created in tandem with radiologists, combines the strengths of humans and tech to make breast cancer screening more effective, more measurable and more accessible. Implementing Vara’s technology increases the productivity of the screening physician while lowering the risk of missing cancer. 

The reality is that every year there are 700k deaths from breast cancer – and 70% of all of these happen in low and middle-income countries. The reason for this is rooted in the ability to detect breast cancer early on, due to a lack of access to technology and specialised doctors. It’s an issue with intersectional implications and one that we should no longer accept. 

It’s reported that catching breast cancer early, through sufficient screening programmes, can significantly increase the chances of survival, from 27% to 98%. But, while only a handful of countries offer such screening programmes, there’s a huge gap in how women experience breast cancer and their chances of survival. This is what fuels Vara –  a desire to drive impact and make a change, making breast cancer screening more widely available and improving the chances of survival for women worldwide. 

We had the chance to talk to the inspiring founder and CEO behind the company, Jonas Muff. Discussing the startup’s mission to make an impact on the lives of women around the world, as well as the company’s unique tech approach and global expansion plans makes for an insightful and emotive interview.

Vara’s mission of making breast cancer screening more accessible is a very laudable goal. What initially prompted you to create a company to try to solve this problem?

Vara began as a project at venture studio Merantix in Berlin, in which people from different backgrounds, all with entrepreneurial mindsets, were brought together to tackle real-world problems in innovative ways. We assembled a team of machine learning experts, software developers, product designers, entrepreneurs and radiologists, and set out to reimagine the breast cancer screening workflow from the bottom up. 

We knew that for the technology to make a difference in the real world – to the workflows of radiologists, and the experience of women being screened – it needed to be embedded throughout the process. With this in mind, we partnered with some of Germany’s leading screening radiologists to build a platform that standardises the entire clinical workflow and enhances it through the use of advanced AI, automation and data management tools. 

What are the barriers that currently prevent effective screening programmes from being established? What can be done to overcome these?

We see that the issues preventing countries from offering routine breast cancer screening programs are complex and multifaceted, and can’t be overcome by just donating scanners, or running software. Through our work, we realised that although breast cancer screening is standard in most European countries, including our home country of Germany, most countries in the world do not offer at-risk women screening. 

This showed us that not only did Vara have purpose, but that there was a desperate and growing need for a platform like ours on a global scale. 

Vara takes an AI based approach to improve the breast cancer screening process. Can you share more about how your product works in practice?

We developed the Decision Referral Pathway, a screening process in which the algorithm only makes a statement for cases when it is confidently making accurate predictions — while leaving other cases to human expertise. 

The goal of the Decision Referral Pathway is to support the radiologist with AI to improve both the sensitivity as well as the specificity i.e. to reduce false negatives and false positives. At the same time, AI is not perfect and cannot make 100% correct predictions for all cases. Therefore, the goal of decision referral is to combine the human expertise of radiologists with the technical capabilities of AI today in a bid to improve both. 

The three types of classification are then:

  • Normal triaging: The algorithm selects a subset of cases that it deems normal with high confidence and automatically labels those cases for the radiologist. The goal of normal triaging is to label as many normal cases negative as possible, with minimal misclassification.
  • Safety net: For cases where the AI is very confident that the images are suspicious for cancer, it offers a safety net to prevent missed cancers. Should the radiologist classify one of those cases as negative, the safety net triggers and points the radiologist to a specific region in the image that is suspicious to the AI. The radiologist can then reconsider the decision, potentially catching a cancer that would have otherwise been missed.
  • Unclassified cases: Importantly, the AI does not make a statement for all cases. There are cases that are neither classified as normal (the least suspicious cases), nor is the safety net activated (the most suspicious cases). For those cases, the AI is not confident enough and the decision expertise should come from the radiologist.

We’re still in the early days of healthtech as an industry. What do you believe is the potential for AI, more generally, in healthcare? 

Firstly, AI on its own is not enough: Vara is not just an algorithm, it’s a full-stack platform. It incorporates AI and automation into every step of the breast cancer screening workflow. It consolidates the previously fragmented processes in one place and does so in a way focused on improving efficiency, enabling remote collaboration and democratising access to better screening programmes. 

Secondly, Vara AI takes a complementary, rather than competing approach to radiologists. Other AI solutions have centred on one-size-fits-all models to replace radiologists. Such approaches may bias the reader and contribute to decreases in reader sensitivity. Instead, our Decision Referral Pathway supports radiologists. It pre-screens normal mammograms, pre-fills structured reports, and post-screens mammograms for potentially missed cancers, all with very high confidence. We finetune the way the AI and individual radiologists interact to optimise for joint performance, while enabling screening radiologists to focus their attention on potentially suspicious exams. Over time, AI will likely have the ability to take over parts of the clinical workflow from human experts, but for now it doesn’t. The work of a screening radiologist is tough and nuanced!

Thirdly, the Vara platform captures real-world data in clinical use. Vara was developed with screening radiologists in one of the world’s most established population-based national breast cancer screening programmes in Germany. It has been trained and evaluated on more than 7 million images to date – one of the largest datasets of its kind globally – and our AI is continuously monitored to provide screening radiologists with key performance metrics. For each woman diagnosed via Vara, we follow the patient pathway after screening to assess AI’s impact on population health metrics such as recall rates and biopsy scores. 

You recently announced a partnership with Mitera, a hospital in Greece. How is Vara working with the team there?

The MITERA and Vara partnership is the first step in our shared mission to increase access to breast screening for women across Greece. Its next vision is to put this comprehensive care closer to women across the country; to enable decentralized screening centers in towns, villages, and surrounding islands. Women in these rural regions are currently disadvantaged due to a lack of access to quality-assured breast screening services. It is envisaged that these screening centers will feed into highly specialized diagnostic clinics where further diagnostic work-up and appropriate management will be ensured.

As a regulated industry, healthtech companies face a lot of challenges when entering new markets. How has this affected your decisions about which markets to expand into? 

Vara enters fast-growing, emerging markets where systems aren’t yet in place to provide women with the access to breast screening that every woman deserves. Through its partnerships, Vara and providers are able to offer screening at an affordable price, which unlocks screening for women who couldn’t access it before. It does this by supporting the commercial partner’s demand generation with a one-off market development investment per new center.

Our business model is built to empower healthcare providers to increase the use of currently underutilised scanners and take a share for each woman screened on the Vara platform. Since Vara takes a share for each woman screened, Vara and the partner share the profits from the increased utilisation. Vara then develops its brand in new markets via the strength of our medical brand and our support to generate demand locally.

Market conditions are quite challenging at the moment. How is this impacting Vara and what is your main focus as a business over the next 12 months?

The Vara Decision Referral Pathway is now in use in 30% of all screening units in Germany and, as part of our global mission, we recently launched screening units in Mexico and Greece, in partnership with healthcare providers on the ground in those regions. We will continue to show considerable traction in those markets by building out the number of screening centres, and we also aim to open additional screening centres in at least one more country. Here, we are looking for the right partner in a country with an under-developed breast cancer screening service where we can have the largest impact. 

It sounds like there is a lot happening at Vara! What are you most excited about?

Vara’s AI performance is showing immense promise in terms of reproducibility and generalisation. One recently peer-reviewed paper, published in the European Journal of Radiology, found that Vara’s AI can correctly detect interval breast cancers in prior screening mammograms, including in some cases with no discernible suspicious signs. This is the first study to assess the potential of AI to reduce interval cancers based on a realistic screening setting with strict requirements for acceptable recall and false positive rates, and was based on the largest cohort of interval cancer cases assessed by an algorithm. Another large retrospective study is currently in press at the Lancet Digital Health. We just started Germany’s first-ever prospective study to understand Vara’s real-world impact in the clinical screening routine.

Noteworthy are our collaborations with renowned academic institutions in the US and Europe, such as Memorial Sloan Kettering Cancer Center, University of Cambridge, Karolinska Institute and the Cancer Registry of Norway. These institutions are evaluating Vara’s performance on breast cancer screening on datasets never seen by Vara and completely independent of the company.

We will continue to challenge the status quo of healthcare globally. In the next four years, we aim to launch 1,000 screening centres in countries that currently do not offer screening. Together with partners, this involves scaling our platform further across Latin America while seeking new partnerships with healthcare providers in other geographies. We have, and will continue to deepen our AI’s capabilities by developing it to not only read the current mammography exam, but to also review prior examinations for even more performance gains. This will help us in our long-term vision to catch every deadly breast cancer early. 

Josephine Conneely
Josephine Conneely
Josephine Conneely is an experienced startup operator based in London. She has helped scaled several high growth startups in Traveltech, Fintech and Enterprise SaaS. She is currently building Cohortly.co, a career community for people working in tech.

Most Popular