AI in Oncology: From Hype to Curated Clinical Tools Amid a Data Tsunami

Key points

• Cancer is the most-searched disease area in AI queries across nine countries, reflecting the complexity of oncology.
• Onqo Health's Easy Oncology, used by two-thirds of German oncologists, provides curated treatment recommendations based on a closed knowledge base.
• AI models like CHIEF achieve high accuracy in cancer detection but require clinical trials for real-world efficacy.
• Overreliance on AI can reduce independent diagnostic skills, as seen in colonoscopy studies where adenoma detection rates dropped after AI removal.
• AI tools must address biases in training data and ensure alignment with current clinical guidelines to avoid inaccurate recommendations.

Background

Previous coverage highlighted AI's potential in cancer research, with Arm CEO Rene Haas noting that while AI could eventually help cure cancer, progress is slowed by chip shortages and the complexity of modeling DNA markers. The broader AI capex wave, driven by tech giants, has reshaped spending, but the focus remains on practical applications in healthcare.

How outlets are covering it

The Guardian emphasizes the gap between AI promises and reality, questioning where cures are despite big tech claims. Konsulteer highlights Onqo Health's curated AI approach, contrasting it with generic tools like ChatGPT, which lack oncology-specific curation. OncLive focuses on AI's role in explaining unexpected trial outcomes, while The Rogersville Review underscores cancer's dominance in AI queries and the need for human oversight to mitigate risks like bias and overreliance.

Why it matters

AI's integration into oncology is critical for managing the growing complexity of cancer care, but its effectiveness depends on curated, domain-specific tools that ensure accuracy and safety. Balancing AI's potential with human oversight is essential to avoid errors and improve patient outcomes.

What to watch

Onqo Health plans to pilot its Board Support product in Offenbach and pursue medical device certification. Continued research will focus on addressing AI biases and ensuring alignment with clinical guidelines, while AI tools will likely expand in cancer detection and treatment planning.