Setting the file. One moment.
Skill 29 · Knowledge Base For Startups
Subchapter 29.209
references/learn/general/mendel-reimagines-how-healthcare-workers-interface-with-patient-data-on-aws.mdMarkdown14 KBView on GitHub
Mendel is using AI and Amazon Web Services to overcome the shortcomings of traditional healthcare tech that limits clinical curiosity.
Whenever we find ourselves sat across from a doctor or nurse, we want to know that they have access to whatever they need to administer the highest level of care. In today’s world, that means data. Every note taken, medication prescribed, or test result received while caring for a patient is part of an evolving story that is unique to them. Healthcare professionals need access to both individualized data and broader contextual data to make informed decisions about the treatment we receive.
Mendel (opens in a new tab) is an innovative artificial intelligence (AI) startup committed to making that process faster than ever before. Together with Amazon Web Services (AWS) (opens in a new tab), Mendel is using AI to overcome the shortcomings of traditional healthcare technologies and processes that limit clinical curiosity. Mendel’s AI deciphers clinical data with clinician-like logic by indexing structured and unstructured data and enabling healthcare workers to interface or chat with patient data using natural language.
Mendel was co-founded by Karim Galil and Wael Salloum in 2017, kickstarting their journey in a two-person office in Silicon Valley. Five years and millions in R&D later, Mendel’s first model was launched. Today, the company is the most-funded clinical natural language processing (NLP) disruptor of its kind, with a team of over 100 members and some of the world’s leading life sciences companies on its client list.
“The biggest problem that healthcare organizations have today is that, while we have a lot of data, most of it is unstructured and isn’t searchable,” says Galil. “Imagine the internet without search engines, that’s exactly what medical records are like today.” Most health care systems utilize multiple electronic medical records (EMRs), each of which contain pieces of the overall patient journey. “Mendel puts every piece of the puzzle together, turning siloed and unstructured data into a single, coherent patient journey.”
“Mendel makes sense of that data and enables healthcare professionals to essentially chat with those medical records,” says Galil. “We have an AI co-pilot store like your typical app store that enables customers to access AI co-pilots for different workflows. For example, we have a co-pilot for file matching, another for prior authorization, and one for abstraction. We also enable our customers to build their own AI co-pilots.” By equipping healthcare professionals with AI-powered tools, Mendel is working to overcome the limitations of existing systems and processes that directly impact patient care.
Today, if a physician wants to manually review a patient chart, the process can take anywhere from 45 minutes to a couple of hours. “If you want to do that at scale for hundreds of thousands of patients, it’s just not humanly possible,” says Galil. “Say a physician wants to compare how effective different drugs have been for patients displaying specific symptoms in the last year, they have to go to a data analytics team who then put together a query.” However, while the analytics team understands how to program code, they lack the nuanced medical expertise of the physician.
“They don’t understand what it means to have colon cancer, for example. And while the physician understands the nuances of medicine, they don’t understand the syntax of the code.” Mendel’s platform overcomes this challenge by enabling end users to interact with data using natural language. “There is no need to write 200 lines of code for a question that can be asked in a couple of sentences in plain English,” says Galil.
“Mendel’s mission is to unravel the truth from patient data at scale. We want to learn from the journeys of hundreds of millions of patients. There’s a lot of disparity in healthcare, where most of the data that we’re leveraging today is from certain ethnicities, certain geographies, certain demographics—we want to change that. We want to learn from patients at scale.”
While the deep learning techniques behind large language models (LLMs) power our most innovative NLP applications, they are limited in enterprise clinical contexts. They enable impressive conversational capabilities but often lack deep contextual understanding.
Galil explains: “Patient records might be written in English, but it’s not plain English, it’s very nuanced language. For example, fatigue can be both a symptom and a side effect.” This can result in superficial reasoning and hallucinations that could have serious consequences in clinical settings.
Rules-based systems also come up short. Rather than providing a holistic understanding of patient care, these systems are unable to replicate the nuanced judgement and adaptive learning of human clinicians. This rigidity doesn’t align with the evolving nature of medical knowledge and practice.
Mendel has developed a hybrid approach capable of clinical reasoning that combines both deep learning and symbolic AI techniques. “We want AI systems that are not only capable of reading, but also capable of reasoning and mimicking how a physician would understand clinical language—that’s the holy grail of AI in medicine today.”
Mendel’s platform comprises clinical reasoning models responsible for serving complaint intelligence from millions of patients combined with leading NLP models that fulfil multiple actions. That includes optical character recognition (OCR), document segmentation, document classification, named-entity extraction, relation extraction, de-identification, and LLM-based retrieval and generation.
Reimagining how healthcare professionals interface with patient data means working collaboratively with different businesses and partners. To that end, finding the right cloud provider is crucial for any AI startup working to achieve their goals as quickly and cost-effectively as possible—which is why Mendel recently migrated to AWS.
“When it comes to AWS, I cannot call them a vendor. They’re more of a partner—that’s what they bring to the table,” says Galil. “It’s one thing to purchase some compute, but it’s a completely different thing to have a partner that understands what your company’s trying to achieve and is willing to get the right people around the table work out how best to execute your cloud strategy.”
Mendel is taking advantage of the AWS Migration Acceleration Program (MAP) (opens in a new tab) to streamline and simplify the migration process. “AWS MAP is a comprehensive program based on best practices and learnings from thousands of successful cloud migrations,” says Christine Chao, Solution Architect, AWS. “It essentially helps businesses migrate faster and with less risk.” The program provides tools that reduce costs and accelerate results, alongside tailored training content, expertise from the AWS Partner Network (APN) (opens in a new tab), and AWS investment.
“One of the first steps in any migration is looking at the data and analyzing the current state of things. We can then make recommendations on how best to proceed. Startups like to move quickly, but taking the time to define an air-tight migration strategy is key to maintaining momentum in the long run,” says Chao. “There are different ways to migrate, some follow a lift-and-shift approach, others are more involved with things like re-platforming. Either way, moving from one cloud to another almost always presents opportunities to optimize. Our job is to help startups seize those opportunities.”
“Part of that process is connecting startups with the right expertise for their use case,” says Chao. “AWS is a big company; we have a lot of industry-specific knowledge to call on and a mature partner network of qualified specialists ready to help.” Galil adds: “A big thing for us was the strength of the AWS Healthcare (opens in a new tab) team which includes physicians who really understand what we’re working on. It’s so refreshing to sit down with someone who can speak that same language.”
“Mendel obviously works with a lot of big tech companies for different aspects of our business. But what is so refreshing about our relationship with AWS is that we feel like we’re dealing with a startup rather than with a big tech company. People are capable of making quick decisions without having to wait for ten different approvals. When you are a startup, you want to work at startup speed rather than big tech speed and we got that with AWS.”
“The AWS Startups (opens in a new tab) team is made up of people that have worked in startups, including founders and investors,” says Chao. “We essentially operate within a startup culture to match our customers’ way of working. That means being available. We’re always in contact with the Mendel team, be that through a dedicated Slack channel or regular meetings.”
The Mendel team are currently working on pre-training a foundation model capable of generating clinical oncology. “It’s a pretty ambitious project,” says Galil. “And when we approached AWS with that, they brought in the right team to advise on what type of compute resources and infrastructure we’d need. Their support went all the way to financially backing the project by offering free credits.”
“As a team, we live and breathe experimentation, which essentially means rapid, scalable, fast iterations,” says Galil. “AWS act as an extension of that team, rather than a separate company. We recently attended an AWS Summit (opens in a new tab), and one of my team members was brainstorming with an AWS account executive around how we can go to market together, how we can develop certain products. Sometimes I get confused—are they on the AWS team or on the Mendel team? That’s what you need from a cloud partner.”
Beyond expertise, Mendel needed a cloud partner capable of securely storing patient data in full compliance with regulatory standards and regional laws like The Health Insurance Portability and Accountability Act (HIPAA). “Privacy is a big deal in healthcare. How information is being stored and handled is a priority for any healthcare startup. A lot of our customers are already on AWS, which makes it easier for us to be there too. We’re meeting them where they are versus us asking them to send data from one cloud vendor to another.”
AWS helps startups develop and evolve their security posture with proven best practices and cutting-edge technology. “Security is job number zero,” says Chao. “We operate under a shared responsibility model where AWS is responsible for the security of the cloud, and our customers are responsible for securing their data in the cloud.” She continues: “We also provide hands-on guidance and resources on how to remain protected and compliant.”
The AI space is hotly competitive, and startups work hard to win a seat at the table. Mendel wanted a cloud partner that could provide the expertise and technology needed to continually deliver value to customers and break new ground in how AI is used in healthcare. Galil explains: “On AWS, you can build whatever you want, and they are there to provide support and push the needle, not compete with you.”
The Mendel team is currently working towards joining the AWS Marketplace (opens in a new tab) with a view to growing its customer base even further. Chao explains: “Getting on the AWS Marketplace enables startups to get their solutions in front of hundreds of thousands of potential customers, build credibility, and close deals significantly faster—there’s even opportunities to co-sell with AWS.”
Every business that joins the AWS Marketplace undergoes a Foundational Technical Review (FTR) (opens in a new tab) to validate their product based on the AWS Well-Architected Framework (opens in a new tab). Once the FTR has been completed, startups gain access to funding, marketing resources, and a Qualified Software badge which immediately signals to customers browsing the marketplace that their products are secure and validated by AWS.
Going forward, Mendel will continue to empower healthcare professionals and improve patient care with AI on AWS. The company is already exploring the potential of AWS Trainium (opens in a new tab) and AWS Inferentia (opens in a new tab) chips as it looks to improve performance and reduce costs associated with its AI workloads. “We are at the frontier of research around AI reasoning in the domain of medicine,” says Galil. “We want to mimic the cognitive abilities of a physician within a machine, and that’s not an easy task.”
“Our partnership with AWS is enabling diagnostic companies and healthcare providers to wield the power of precision medicine with a click of a button. The future of medicine is going to be a collaboration between machines and humans. When companies like AWS work with companies like Mendel, that future becomes now rather than a few decades from today.”
Author: AWS Editorial Team
The AWS Startups Content Marketing Team collaborates with startups of all sizes and across all sectors to deliver exceptional content that educates, entertains, and inspires.