Generative AI for the Healthcare Sector

We have a great health system, but it’s not perfect. A few challenges include: 

• Delivering services despite workforce shortages 
• Adapting the system to meet the challenges of an ageing population and chronic disease 
• Ensuring equitable access for everyone to healthcare services  

Technology has an important role to play (virtual care, electronic records, genomics, monitoring devices). AI in health has been around for decades. I personally used it 24 years ago and now it’s in widespread use. 

However, prior generations of machine learning have been expensive and time consuming to train resulting in specific models with narrow focuses. The significant change with generative AI, like OpenAI’s ChatGPT, is that it’s pretrained on a vast amount of global data. Applying AI to different use cases is now the primary focus, rather than training it from scratch.  

OpenAI and Microsoft are two different companies, with many similar but some different views. We have recently announced the third phase in the partnership that amounts to a multi-year, multi-billion-dollar investment. 

My focus at Microsoft is delivering the GPT foundation models wrapped as an enterprise service within Microsoft’s Responsible AI framework. 

Our responsible AI team has hundreds of people providing governance ahead of regulation and legislation. We prioritise fairness, reliability and safety, privacy and security, inclusiveness, transparency, and accountability.

I am working with a cross-ANZ selection of large health providers to explore use cases. We have found that generative AI can have value in the following ways: 

• Decreasing the administrative burden of clinical documentation 
• Increasing the completeness and quality of clinical notes 
• Translate and summarise complex information and research 
• Do mapping between free form narrative and structured vocabularies 
• Powerful semantic search with citations to check credibility and accuracy 
• Make information available to patients succinctly and in context (bots) 
• It can even write code 

Clinicians are generally amazed and hopeful about the possibilities of AI. They recognise that this technology will change their roles, but it will create new skills and jobs. Most importantly it allows clinicians more time with the patient, doing the higher order functions. It’s letting doctors be doctors. 

The integration of AI in Healthcare is happening in these ways: 

1. Health organisations directly exploring the capability of AI 
2. Software companies making capabilities available in their products with tightly defined capabilities: Epic, Nuance, Opus5k 

We are optimistic that this technology will have a huge impact on healthcare. We are excited to collaborate with governments, healthcare providers, industry groups and researchers to explore all the possibilities.