Project DORIS: Pioneering Precision Medicine Information Access with AI Innovation
Data, AI, generative language models are fast becoming ubiquitous in venture investment. With global investment in generative AI sitting well above US$ 14 billion in 2023.[1] The vast majority of this investment to date has been into foundational models (GPT3, GPT4 and now GPT 5) and horizontal infrastructure (e.g. Chat GPT, Google’s Bard) and at some incredibly high valuations.
Compared to this, investment in vertically specific generative AI applications, leveraging industry know how and experience, is still very nascent, with a lot of opportunity to create companies that can harness the innovation in AI and LLMs in combination with other proprietary elements. A customized and enriched user experience coupled with proprietary data to train a model for high accuracy and essential domain-specific expertise can be used create platforms to do something that no one else’s model can, and to solve very real problems.
In August, Bridgewest Ventures partnered with Wild Type Advocates to invest in Project Doris – a new innovation that utilises AI and LLMS to change the way clinicians access and make use of cancer treatment information. The unique thing about this team and technology is their deep understanding of their users and the way in which they approach the build of this technology from the perspective of the clinicans they are trying to serve. In this blog post, we talk a little bit about why we invested.
Information overload vs. limited treatment time
Over the past decade, annual approvals of cancer therapies have quadrupled, leaving healthcare providers overwhelmed and struggling to keep up with constantly updated clinical guidelines and drug labels. On average, physicians’ treatment time is limited to 20 minutes/month per patient, while clinicians realistically need up to 5 hours to navigate through hundreds of dense guideline pages to build a treatment plan for a complex case. This leads to growing disparities of care, increasing errors in treatment decisions, and reduced access to potentially life-saving, state-of-the-art treatments. Patients in rural areas, who have a demonstrably higher mortality rate than those from more central regions, are even more affected by this problem.[2]
"NCNN guidelines are an invaluable resource, but the pdf flowchart format is outdated and difficult to follow.” – Chicago Oncologist KOL
“85% of Oncologists are interested in a GPT-like chat for oncology.” – Wild Type Advocates insights research
In their consultancy practice, the founding team became interested in this problem and ways in which technology could help solve it. Through their initial research, they found that 100% of treating oncologists (n=44) under 40 years old have tried ChatGPT but were all disappointed in the information provided, and they are crying out for easier, faster and better access to the latest medical resources.
DORIS, your Diagnostics and Treatment Reference Librarian
In a leap towards revolutionizing medical care and improving patient outcomes, DORIS – Dynamic Oncology Reference Information System – emerges as the beacon of Precision Medicine. A fine-tuned, chat-driven information platform trained on current Oncology data with timely updates. DORIS uses a set of proprietary ingestion tools, quality methods and enriched interface to create a reference librarian agent who provides doctors and patients fast access to accurate cancer diagnostic and treatment information.
Imagine a world where thousands for pages of medical treatment guidelines, drug labels, clinical outcomes and clinical trials data are captured in a medical and diagnostic knowledgebase that is instantly accessible, putting the most relevant information that you need right at your fingertips. That is the mission of Project DORIS and to achieve it, the DORIS technical team have to break a few rules in AI and machine learning, inventing new methods to digest and incorporate different data types and data sets into a highly accurate tool, minimizing common pitfals of AI models to build a product users can trust.
“We want to make Precision Medicine work for everybody.”
- Joe Monforte, Founder Project DORIS
DORIS paves the way for this transformative change by harnessing the power of data and technology to provide personalised, up-to-date medical solutions so the clinician is best placed to make well-informed clinical decisions.
What sets DORIS apart
Their customer centricity makes the difference. “Instead of being tech outsiders, we have all worked within the cancer community for decades. It’s a product built by people who work in oncology for people who work in oncology.”
The Project DORIS team is highly insights and customer driven. “Just within the last three months, we have talked to over 300 doctors who continuously give us feedback on their needs.”
This deep understanding of their users and the environment within which they operate drew the Bridgewest team into working with them to build a company that goes beyond the technology used to date to solve huge challenges in data digestion and interpretation.
Complementing team forces
The driving force behind the DORIS Joint Venture is a visionary team of founders with exceptional expertise in both medical and technological domains and an obsession with solving physician burnout and healthcare equity problems globally.
Led by Dr. Joe Monforte, a seasoned life sciences entrepreneur; Dr. Pia Gargiulo, a pharmaceutical and business development executive; and Mr. David Holecek, a healthcare executive and commercialization consultant, the team combines diverse skills, in-depth industry knowledge and extensive networks. Complementing this trio is Mr. Dion Cummings, ex-IBM Watson and a distinguished data scientist with expertise in synthesis AI.
Thriving in New Zealand's Healthcare Innovation Hub
“We are excited to prove this in NZ” - Joe Monforte, Founder Project DORIS
Project DORIS recognizes the exceptional value to advancing its ground-breaking healthcare initiatives in New Zealand. By actively engaging with New Zealand's oncologists during beta testing, DORIS will gain insights into tailoring its technology to suit diverse healthcare perspectives internationally. The founder Joe Monforte highlights: “It is wonderful test lab. You try it and see what works in a population of five million before you try it in a population of 330 Million.”
The decision to base the company in New Zealand underscores the project's commitment to nurturing local expertise, contributing to the growth of the nation's tech and healthcare sectors.
The Project DORIS team plans to fortify data management and medical decision-making by leveraging expertise within the New Zealand healthcare community, expanding globally, and actively participating in AI/LLM-focused groups such as the AI Forum NZ LLM Working Group to advance healthcare innovation.
What happens next
“We want to get the product into hands of oncologists, get them trying it, learn from their usage and further improve it.”
This funding round will focus on developing out the DORIS prototype for use in 4 cancer types. The goal will be to create a product that goes where other LLMs have not gone before, delivering clinician validated, highly accurate answers that are 100% grounded in validated clinical guidelines and label information. Behind the scene, the build team and AI experts are having to develop new methods of ingesting and processing highly complex data sets.
“Our MVP, which we just finished, where we ingested all the core materials for the diagnosis and treatment of breast cancer is just the start.”
With the money from the founding round the team is focused on developing the MVP, a prototype of DORIS, involving a chat UI powered by LLM and a select set of data sources. This phase necessitates creating a novel set of backend tools for data aggregation and auto-curation. The outlined Technology Development plan spans from project definition and data collection to software architecture and UX/UI design, and will culminate in a chat-driven product, developed, tested for accuracy, and deployed for the first users to try.
“We are working now on building DORIS for 4 large cancer types and that is going to be our alpha that we will launch in March.”
By incubation's end, a functional prototype will undergo trials with oncologist partners. Early performance assessments will guide post-incubation technology refinement, aiming for enhanced accuracy and usability. Further development post-incubation precedes the tool's market launch and widespread adoption.
“We will do an expanded access to the tool to oncology healthcare providers to get feedback from them on how they want to do it and what kind of work flows do we want to enable and make most efficient within the tool.”
[1] PitchBook • Geography: Global • *As of June 30, 2023
[2] Nixon, G., Davie, G., Whitehead, J., Miller, R., de Graaf, B., Lawrenson, R., ... & Crengle, S. (2023). Comparison of urban and rural mortality rates across the lifespan in Aotearoa/New Zealand: a population-level study. J Epidemiol Community Health.