Navigating AI

Date:  24 June 2024

On 18 June 2024 the Big Data Special Interest Group, led by SIG sponsor Vizient, hosted its latest meeting on “Navigating AI”. The Big Data SIG, sponsored by Vizient, connects experts in the application of data for decision-making in hospitals and health systems.  

The session, chaired by David Levine from Vizient, began with an introduction to equity and bias in AI, under the umbrella of governance. David was joined by guest speaker Andrew Rebhan, lead analyst for health data intelligence at Sg2/Vizient.  

Addressing equity and bias in AI deployment 

Andrew Rebhan delivered an insightful presentation on AI bias, its risks and challenges and mitigation strategies. He traced the journey from the initial waves of excitement around ChatGPT to a shift to concerns about AI’s potential risks and challenges, particularly bias and its broader impact in society. 

AI bias within healthcare exists when the application of an algorithm compounds existing inequities, and amplifies them to adversely impact inequities in health systems. Causes of bias include  structural inequities, incomplete health data across populations, subjective interpretation of AI input and faulty algorithm design or problem framing. 

Besides this, there are some additional challenges such as the fact that AI transparency isn’t easily quantified or that bias isn’t always obvious. 

Strategies to mitigate bias span the entire AI development cycle 

Andrew highlighted how strategies to mitigate bias need to be deployed during the entire development cycle, and leaders must be proactive in calling out bias during: 

  • Problem framing 
  • Data sourcing and processing 
  • Model development and validation 
  • Deployment and integration 
  • Monitoring and maintenance 

Other challenges include how we develop standards at each of these stages that promote safety, data privacy and equity, and how we provide ongoing education and training for stakeholders on these challenges and risks. 

AI has a role in promoting health equity 

The meeting highlighted that AI has enormous potential to promote health equity by personalizing care, considering socio-economic, cultural, and genetic factors, and improving accessibility . However, there is work to be done by the industry to tackle the challenges outlined here, whether that is bias, equity or cybersecurity.  Finally, Andrew stressed that a sustainable AI programme requires building the strategic and technical foundation, anticipating common barriers, prioritizing trust and change management, finding strategic partners to fill in the gaps and keeping an eye on reskilling the workforce. 

In his closing words Andrew affirmed that AI bias can be addressed, and that these challenges are not insurmountable. But tracking bias is a problem that requires ongoing monitoring. Find the whole presentation here!

News & insights

Look into our latest...

Navigating AI

On 18 June 2024 the Big Data Special Interest Group, led by SIG sponsor Vizient, hosted its latest meeting on...

International hospital collaboration (IHC): how can we promote ethical and responsible partnerships?

Since the 1970s, hospital community practices have undergone profound changes due to evolving public health issues such as epidemiological transitions...

IHF members at the WHA: Strengthening global connections in Geneva

In May 2024, the 77th World Health Assembly brought together global health leaders. As a non-state actor in official relation...

Key sessions at the 77th World Health Assembly

In May 2024, the world tuned in to the 77th World Health Assembly (WHA) themed “All for Health, Health for...