Xiao Liu

This research is just the first step on a long, exciting journey. We’re excited to keep testing and learning with our healthcare partners in pursuit of better, more accessible care for people everywhere. More on the blog today: microsoft.ai/new/the-path...

MAI-DxO in action, tackling one of those complex cases:

Agentic A.I. vs experienced physicians for diagnosis of > 300 complex diagnostic cases:
4-fold higher accuracy and 20% lower cost arxiv.org/abs/2506.22405

We're taking a big step towards medical superintelligence. AI models have aced multiple choice medical exams – but real patients don’t come with ABC answer options. Now MAI-DxO can solve some of the world’s toughest open-ended cases with higher accuracy and lower costs.

AI in healthcare: what does good evidence and regulation look like? Artificial intelligence is infiltrating medicine at many levels, from providing patients with medical advice and summarising patients’ notes to aiding diagnoses. Elisabeth Mahase looks at how the rese...

AI tools have a lot of potential to improve medicine and the healthcare landscape.

But how are they regulated? And what does 'gold standard' evidence look like?

@jessrmorley.bsky.social @xiaoliu.bsky.social @bmj.com
www.bmj.com/content/389/...

Our team at @unibirmingham.bsky.social have supported the UK National Screening Committee in the evidence approach for #AI in Diabetic Eye Screening over last few years.

In this paper we outline the NSC's considerations for evaluating ARIAs in England.

doi.org/10.1016/j.la...

Great to be at NICE Conference yesterday for a closing discussion on AI! Can we ensure evidence generation is as adaptive and intelligent as the technology, without compromising definitions of 'Safe, Effective and Equitable & do it in a way that is collaborative, transparent and trustworthy?

The recording of this talk on STANDING Together is available if you missed it!

www.datadiversity.org

@ai.nejm.org is now on bluesky! 🥳

📣 CANAIRI: the Collaboration for Translational AI Trials! Co lead @xiaoliu.bsky.social @naturemedicine.bsky.social

Perhaps most important to AI translation is the local silent trial. Ethically, and from an evidentiary perspective, this is essential!

url.au.m.mimecastprotect.com/s/pQSsClx14m...

CANAIRI: the Collaboration for Translational Artificial Intelligence Trials in healthcare - Nature Medicine Nature Medicine - CANAIRI: the Collaboration for Translational Artificial Intelligence Trials in healthcare

CANAIRI - Collaboration for AI Translational Trials - takes flight!🪽
We will work together to define how to take a #sociotechnical approach to responsible #AI integration in healthcare. Spearheaded by my awesome co-lead @mdmccradden.bsky.social 🦜
nature.com/articles/s41...

Tackling algorithmic bias and promoting transparency in health datasets: the STANDING Together consensus recommendations Without careful dissection of the ways in which biases can be encoded into artificial intelligence (AI) health technologies, there is a risk of perpetuating existing health inequalities at scale. One major source of bias is the data that underpins such technologies. The STANDING Together recommendations aim to encourage transparency regarding limitations of health datasets and proactive evaluation of their effect across population groups. Draft recommendation items were informed by a systematic review and stakeholder survey.

Hello there BlueSky, it's a great pleasure to be able to share a project our team has been working on these last three years!

(Led by the venerable @xiaoliu.bsky.social)

www.thelancet.com/journals/lan...

1/

Tackling Algorithmic Bias and Promoting Transparency in Health Datasets: The STANDING Together Consensus Recommendations Without careful dissection of the ways in which biases can be encoded into artificial intelligence (AI) health technologies, there is a risk of perpetuating existing health inequalities at scale. O...

Published in The Lancet Digital Health and NEJM AI, and a core part of this project - the patient and public's perspective in Nature Medicine:
tinyurl.com/standingnejm...
www.nature.com/articles/s41...

YouTube video by University of Birmingham Founders' Awards 2024 - The Standing Together team | University of Birmingham

youtu.be/dVpSh_7NpR8

Tackling algorithmic bias and promoting transparency in health datasets: the STANDING Together consensus recommendations Without careful dissection of the ways in which biases can be encoded into artificial intelligence (AI) health technologies, there is a risk of perpetuating existing health inequalities at scale. One major source of bias is the data that underpins such technologies. The STANDING Together recommendations aim to encourage transparency regarding limitations of health datasets and proactive evaluation of their effect across population groups. Draft recommendation items were informed by a systematic review and stakeholder survey.

One of my biggest joys is seeing an entire community take a stand for something important.
STANDING Together, recommendations for health data diversity and algorithmic bias is now published: an international effort involving >350 experts & 58 countries.
www.thelancet.com/journals/lan...

UK government failing to list use of AI on mandatory register Technology secretary admits government departments are not being transparent over way they use AI and algorithms

Only 9 algorithms submitted to the UK Gov's #AI public register so far. Including just 3 since 2022.

MHRA trials five innovative AI technologies as part of pilot scheme to change regulatory approach The pilot scheme, AI Airlock, is designed to help test and improve the rules for AI-powered medical devices to ensure they reach patients quickly, safely and effectively.

www.gov.uk/government/n...

The MHRA has just announced 5 AI products for its Airlock (Regulatory Sandbox) programme. I’m on the governance board and am excited to see this take flight.

The products include:
- two LLM based devices
- a platform
- a personalised care product (breast cancer)
- a risk prediction product (COPD)

Shannon Vallor says AI does present an existential risk — but not the one you think AI is not a “stochastic parrot,” It’s a mirror, says the philosopher.

“It’s the rhetoric of AI today that is about gaslighting humans into surrendering their own power and their own confidence in their agency and freedom. That’s the existential threat, because that’s what will enable humans to feel like we can just take our hands off the wheel and let AI drive.”

The #SAIL25 CALL FOR ABSTRACTS is now open! Join SAIL 2025 on May 6-9 in Río Grande, Puerto Rico! In-person attendance is limited to those with accepted work.
NEJM_AI will award cash scholarships to the top abstracts. Travel support is also available. Submit here by January 17:

Diversity, inclusivity and traceability of mammography datasets used in development of Artificial Intelligence technologies: a systematic review There are many radiological datasets for breast cancer, some which have supported the development of AI medical devices for breast cancer screening and image classification. This review aims to identi...

First up (led by Dr El Laws):

www.clinicalimaging.org/article/S089...

We show gaps in the #data landscape for #mammography, with demographic data often missing, and the entire continent of Africa excluded.

This means the #datasets currently available do not represent most of the world.

Hello Bluesky! It seems nicer here than in the other place 🥲

I'm Joe: a medical doctor & clinical researcher based in Birmingham, UK. I work with @xiaoliu.bsky.social & co. on all things #AI in #healthcare.

We have a few papers coming out in the coming weeks. Below are a few from earlier this year

Excited to be speaking at #THISspace2024 tomorrow on Responsible data science and health innovation, hosted by @thisinstitute.bsky.social

This MSc is for anyone from a clinical, technical, or operational background. It will prepare you to lead multidisciplinary teams to deliver AI-enabled benefits to patients and services.

Artificial Intelligence Implementation (Healthcare) MSc / PGDip - University of Birmingham There is an urgent need for leaders of AI implementation across the healthcare workforce of the UK and worldwide. The MSc AI Implementation (Healthcare) programme is designed to address this need.

We are launching a new MSc in #AI Implementation (Healthcare) programme at the University of Birmingham (UK) in 2025

Data with no or almost no documentation, being forced to make a bunch of assumptions about how the data came about, then realising they probably weren’t true later

A knotted pipeline Data-driven systems and inequalities in health and social care

The Sudlow Review resonates with Ada's research with @healthfoundation.bsky.social. The data 'pipeline' underpinning data-driven systems in healthcare is full of ‘knots’ which disrupt how well they work, creating systemic and unequal impacts on people. www.adalovelaceinstitute.org/report/knott...

This is an excellent starter pack for anyone in #Medsky looking to follow all the national bodies who have now joined Bluesky.

go.bsky.app/a2U8nw

Gaps in health #data drives health #inequalities in an #AI and data-driven future. We wrote about this in NEJM AI and are working on it at STANDING Together: www.datadiversity.org