Google Develops AI for Ultrasound Diagnosis and Cancer Therapy

At its annual healthcare event, The Check Up, Google announced partnerships with medical organizations to explore the use of AI in ultrasound readings, medical language models, and cancer treatments. The company sees AI as a critical tool in simplifying ultrasound diagnosis in areas with a shortage of trained specialists. Google’s AI models could help identify data such as early breast cancer detection and gestational age in pregnant women. The company is partnering with Jacaranda Health in Kenya to research AI-based ultrasound treatments for mothers and babies in government hospitals.

Google is also collaborating with Chang Gung Memorial Hospital in Taiwan to explore how AI can detect breast cancer via ultrasound as an alternative to mammograms, which are less accessible in lower-resource regions. The company’s next-generation healthcare large language model (LLM), Med-PaLM 2, has improved significantly, scoring 85% on doctor-level medical exam questions – an 18% increase from the previous version.

Additionally, Google has partnered with Mayo Clinic to study how AI can help with radiotherapy planning for cancer treatment, reducing the time-consuming process of “contouring” and drawing lines on CT scans. Finally, the company is working with an AI-based organization to provide AI-powered tuberculosis screenings in Sub-Saharan Africa, with 100,000 free screenings donated to help detect the disease early and provide early treatment to reduce its spread. Despite these developments, Google cautions that the technology is not yet ready for real-world work settings, and significant gaps in scientific factuality, precision, and bias remain.