Revolutionizing Pediatric Rare Disease Diagnosis: ThinkRare AI Goes National
In a landmark advancement for Canadian healthcare, the CHEO Research Institute has announced the national expansion of its groundbreaking AI algorithm, ThinkRare. Developed specifically for pediatric patients, this world-first tool is set to transform how clinicians across the country identify and diagnose undiagnosed rare genetic diseases (RGDs), often referred to as rare diseases. With rare diseases affecting approximately one in 12 Canadians—equating to over one million individuals, 75% of whom are children—this expansion couldn't come at a more critical time.
ThinkRare represents a pinnacle of responsible AI integration in clinical settings, scanning electronic medical records (EMRs) in real-time to flag potential cases that might otherwise evade detection for years. By alerting care teams proactively, it shortens the infamous 'diagnostic odyssey'—the average 5 to 7 years families endure seeking answers—and paves the way for timely genetic testing and intervention.
The Burden of Rare Diseases in Canadian Pediatrics
Rare diseases, defined as conditions affecting fewer than 1 in 2,000 people, encompass over 7,000 known disorders, most with genetic origins. In Canada, pediatric cases dominate, with 900,000 children impacted. Shockingly, one-third remain undiagnosed despite exhaustive efforts, leading to misdiagnoses in 40% of cases and prolonged uncertainty for families.
Symptoms often overlap with common conditions—developmental delays, unexplained seizures, or growth issues—resulting in referrals across multiple specialists. This fragmented care burdens families emotionally and financially while straining healthcare resources. Initiatives like CARE4Rare Canada, led by CHEO researchers, have diagnosed hundreds through genome sequencing, but scaling remains challenging without tools like ThinkRare.
- 25% of patients wait 5-30 years for diagnosis.
- Half never receive one.
- Genomic testing solves ~30-40% of cases when applied early.
For parents, the journey involves endless appointments, tests, and hope. ThinkRare addresses this by automating pattern recognition in vast EMR data.
How ThinkRare Works: A Step-by-Step Breakdown
Unlike black-box machine learning models, ThinkRare is a transparent, rules-based algorithm—combining clinician-defined criteria with structured EMR data. Here's the process:
- Continuous EMR Scanning: Monitors patient records for predefined flags like multiple specialist visits, negative common tests, and symptom clusters (e.g., hypotonia, failure to thrive).
- Eligibility Check: Confirms age (pediatric), no prior genetics referral, and exome sequencing suitability.
- Alert Generation: Issues a clinician-facing 'Best Practice Advisory' (BPA) with rationale and referral recommendation.
- Human Review: Physicians assess and act, ensuring ethical oversight.
- Feedback Loop: Performance refined via retrospective validation.
In a proof-of-concept on 262,296 CHEO patients, it excluded 99.9%, flagged 30 for sequencing, yielding 60% recall and 15% precision. Prospective use identified 41 referrals, with 70% diagnosing RGDs.
This precision balances sensitivity to catch cases without overwhelming clinicians.
Proven Success: Real Stories from CHEO
Since deployment, ThinkRare's impact is tangible. One standout case: 10-year-old Antony Wistaff, premature with hydrocephalus, endured 100+ appointments across six clinics and 30 ER visits. Flagged by ThinkRare, genome sequencing revealed Chung-Jansen Syndrome (PHIP gene variant), affecting ~400 worldwide. The diagnosis explained delays and enabled targeted support, bringing 'peace' to his family.
Overall, 41 flagged children referred to genetics; 70% diagnosed. This demonstrates ThinkRare's role in ending diagnostic odysseys, published in Genetics in Medicine (Nov 2025).
These outcomes underscore AI's potential in precision medicine.
Photo by Markus Winkler on Unsplash
From Local Tool to National Platform: Expansion Details
Announced February 27, 2026, ThinkRare's rollout begins with McMaster Children’s Hospital (validation complete), followed by Alberta Children’s and Stollery Children’s Hospitals. CHEO's team, aided by HHSC CREATE, adapts for diverse EMRs, ensuring governance, safety, and transparency.
Challenges include system interoperability and ethics, but successes position Canada as an AI-healthcare leader. Future sites eyed nationwide, pending funding.
Read CHEO's official announcementLeadership Driving Change: Key Figures at CHEO
Dr. Kym Boycott, Chief of Genetics and Canada Research Chair in Rare Disease Precision Health, co-leads ThinkRare, envisioning it shortens odysseys. Ivan Terekhov, Director of Research Informatics, AI & Technology, oversees technical scaling. Alexandre White-Brown manages implementation. CEO Dr. Jason Berman champions clinician-data-family collaboration.
"Rare diseases hide in plain sight... Scaling ThinkRare will shorten the diagnostic odyssey." — Dr. Kym Boycott
Their work builds on CARE4Rare's decade of genomics advances.
Implications for Canadian Healthcare and Research
ThinkRare exemplifies AI's role in overburdened systems, reducing costs (genomic testing ~$5-10k vs. years of care) and improving outcomes. Nationally, it could diagnose thousands, integrating with Canada's Precision Health Strategy.
In higher education, it highlights demand for AI-health experts. Explore research jobs or AI career advice to contribute.
| Metric | CHEO Results |
|---|---|
| Patients Scanned | 262,296 |
| Flagged | 30-41 |
| Diagnosis Rate | 70% |
| Recall/Precision | 60%/15% |
Challenges, Ethics, and Future Outlook
Ethical AI is paramount: ThinkRare's transparency (rules-based, auditable) mitigates bias. Challenges: EMR variability, clinician trust, funding. Future: Broader AI tools, multi-omics integration, global export.
Dr. Boycott notes: "My ultimate goal." With partners like McMaster, Canada leads responsible AI.View the Genetics in Medicine study
Photo by Artyom Korshunov on Unsplash
Why This Matters for Families, Clinicians, and Researchers
For families: Answers sooner. Clinicians: Efficient triage. Researchers: Data for genomics. As AI evolves, tools like ThinkRare herald precision pediatrics.
Interested in genetics or AI roles? Check higher-ed jobs, rate your professors, or career advice. Share experiences in comments.