CoeurWay: an AI scribe that reduces medical note writing time by 70%
By CoeurWay | Published on October 3, 2024
Problem and objectives
Problem
- Considerable administrative burden: In Canada, an estimate of 30 million hours devoted annually to non-delegable administrative tasks by physicians.
- Impact on clinical time: Healthcare professionals spend up to 3 hours per day writing notes, not including work done outside of regular hours.
- Strict ethical requirements: Obligation to produce comprehensive clinical notes, in accordance with professional standards.
- Professional burnout: Increased professional burnout, exacerbated by the overload of administrative tasks with low added value.
- Critical shortage of personnel in the health sector: Lack of doctors and many other health professionals in Quebec.
Objectives
- Automated documentation of consultations: Providing healthcare professionals with a tool to automatically document encounters with their patients.
- Significant reduction in administrative tasks: Free professionals from hundreds of millions of hours of administrative tasks.
- Increased capacity to provide care: Allow health professionals to offer more services to the population without compromising their well-being.
- Improved quality of patient interactions: Freeing professionals from the burden of real-time note-taking, enabling more engaged and attentive interactions with patients.
- Burnout mitigation: Help reduce burnout by reducing administrative burden.
- Ergonomic integration into existing practices: Develop an easy-to-use tool that integrates harmoniously into the work process of professionals.
Proposed solution
Solution
CoeurWay is a Quebec startup that has developed artificial intelligence aimed at automating medical notes and forms for all types of health professionals, simply by listening to consultations.
Founded by Marie Lavoie and Antoine Lavoie, two healthcare professionals with three years of experience in medical artificial intelligence in New York at FitCam Health. And Pierre-Olivier Gravel, an expert in data science and LLM. In addition, they were accompanied by rich support from the entrepreneurial community, incubators and accelerators, putting CoeurWay on the right track from the start.
Principle
CoeurWay is a web application that leverages the latest advances in generative AI and speech recognition to generate high-quality medical notes in the usual structure used by doctors in less than 10 seconds. On average, doctors take 10 minutes of writing per note. Doctors using CoeurWay are able to make their notes in less than 3 minutes, which includes checking the generated note. This is an efficiency gain of 70% on the time usually spent on writing.
User experience
Harmonious implementation in the work process of professionals :
- Speech recognition of over 99+ languages
- Specialization in Quebec French and medical terminology
- Note style adapted to each profession: Doctor, Nurse, Psy, Social Work, Physio, etc.
- Virtually no changes to the usual work process
- Very simple web interface
- Technical support team on the lookout for user needs
- Chatbot available to answer frequently asked questions
Generating personalized notes :
- Notes adapted to the reality of each type of profession
- Note style and format as per user preference
Key figures (since June 2024) :
- 500+ active users
- 24,000 notes generated
- 98% acceptance of patient consent for recording
- 2,800 hours of writing saved
- 11,200 potential additional appointments
Role of AI
Voice recognition | Natural language processing:
- Transcription of meetings between health professionals and patients
- Voice recognition adapted to the Quebec accent
- Medical vocabulary and lexicon of medications used in Quebec
Large Language Model | Generative model:
- Generation of notes adapted to several types of healthcare professionals
- Personalization of note generation according to user preferences
- Adaptation of models according to feedback received
Automated Machine Learning (AutoML) | Supervised learning:
- Processing chain (pipeline) allowing continuous improvement of the model
- Generation of anonymized data
Impacts
Main outcomes
Medical teams:
- Gain in efficiency for healthcare professionals
- Free up dozens of hours per month
- Improves the quality, readability and completeness of notes
- Improves job satisfaction for professionals who can focus on clinical tasks rather than documentation
Patient care:
- Healthcare professionals can pay more attention to their patients
- Eye contact without constant interruption, possibility of noting clinical signs, important details or interpreting the non-verbal language of patients
- Possibility of generating a simplified report for patients following the meeting from the data collected
Population health:
- Greater availability of health professionals
- Increases the ability of healthcare professionals to see a larger number of patients
Economic development:
- Growth from 0 to 17 employees between 2023 and 2024
- Objective of reaching 50 employees by the end of 2025
Challenges takled
Performance in compliance: Be efficient in terms of both the quality of notes and speed while being compliant
Offer a solution adapted to Quebec French: Adapt voice recognition to Quebec French and the medical language used in Quebec despite the limited quantity of medical audio data structured in this language
Gain the trust of professionals: A target audience of extremely busy people who have very high standards - One chance, one opportunity
Compliance with the legal framework: Respect the numerous laws and legislation in force in Quebec regarding cybersecurity and the security of health-related data
Data management: Responsible and ethical processing and use of anonymized data
Conditions for success
Team mobilised
Mobilized team
17 full-time employees
Areas of expertise:
- Data Architect
- Expert in LLM
- Software engineer
- Expert in UX/UI
- Business development
- Marketing
- Healthcare professionals
Collaborations
NextAI and Mila
MEDTEQ+
Faculties of Medicine of Quebec universities
MAclinique networks
Unimedic
Funding received
Personal funds:
- Founder: CA$120K
Private funds:
- Angel investor: CA$250K
- Investor doctor: CA$300K
- Round seed $2M CA (coming soon)
Public funds:
- Canadian government fund: CA$291k (CNRC, ÉléverlaPI...)
- Quebec government fund: CA$210K (MEIE)
Project stages
- May 2022: 18 months of research and development - December 2023: 6 months of prototype beta with more than 200 professionals - June 2024: Official launch of the platform - Today: Over 500+ users
Project stages - to come
- October 2024: Finalization of a seed round to accelerate commercialization - December 2024: Template creation module 2025: Product: - Q1 2025: Integration with EMRs, Module for creating personalized note formats by users, Completion of PDF documents (e.g. CNESST, SAAQ, Insurance forms), Community for sharing note templates. - Q2 2025: Collection of information prior to the consultation, Inter-professional collaborative workflow, Solution compatible with telehealth - Q3 2025: Automatic recording of medical information in the patient's file (Allergies, Medications, Laboratory results, etc.) Marketing: - 90% of the Quebec market + AI Scribe solution chosen by Santé Québec
Context of the solution
Customers
The versatility and versatility of CoeurWay allows its deployment to all professionals in the health system, easily adapting to the particularities of each profession with as much precision as flexibility.