Michael Barnes (@thebarneslab) header
Michael Barnes (@thebarneslab) profile picture

Pharmademic Computational Biologist @QMUL_C4TB

London
Joined December 2012
Photo by Michael Barnes (@thebarneslab)Photo by Michael Barnes (@thebarneslab)Photo by Michael Barnes (@thebarneslab)Photo by Michael Barnes (@thebarneslab)Photo by Michael Barnes (@thebarneslab)Photo by Michael Barnes (@thebarneslab)Photo by Michael Barnes (@thebarneslab)Photo by Michael Barnes (@thebarneslab)Photo by Michael Barnes (@thebarneslab)
Just out! Our package for exploring the temporal dynamics of time course data, with a focus on Burstiness of events. @AI_MULTIPLY #AI doi.org/10.48550/arXiv.2411.…
1
3
4
257
We are using our package to characterise the burstiness of events in patient health records to better understand the triggers of poor health in patients with multiple long term conditions part of the #NIHRAIM programme
2
2
171
Predicting perturbation, maybe a handy AI trick for drug discovery?
1/14 Excited to introduce PertEval-scFM! It provides benchmark and evaluation tools for perturbation effect prediction models, including single-cell foundation models (scFMs). Paper and GitHub link at the end of the thread! 🧵👇
3
381
We're really proud to publish our Shiny R Expression Portal, making it easy for others to present any type of expression data on the web. Great work from @rafaelhenkin and team!
3
12
661
Michael Barnes retweeted
UK & Ireland Major Trauma Haemorrhage: What Now? What Next? 12th & 13th March, Wellcome Trust, London & online Register for free here: eventbrite.co.uk/e/uk-major-… Critical bleeding is still the leading cause of potentially preventable deaths in major trauma patients. Despite huge improvements in overall outcomes over the past 10 years, mortality from torso haemorrhage remains very high, and over 50% in some series. This is a national (UK & Ireland) meeting to bring together all MTCs, Air ambulances and anyone interested to discuss future management, pathways and prrotocols for bleeding management. (Worldwide attendees online are very welcome) The past year has seen two major clinical trials completed across the national trauma system – UK-REBOA and CRYOSTAT2. What does data from these trials tell us about our current pathways and practices for major haemorrhage. What should happen now? What are the implications of these studies? How should we change our practice? And can we look to the future to improve our decisions, practices and outcomes? The aim of the event is to use results from the trials and other contemporaneous data to facilitate a national discussion regarding how they should be incorporated into practice in the UK. We will cover the bleeding pathway from prehospital care through to operative or interventional haemorrhage control. The event will be fully interactive, and we would really like to bring out the views and experiences of front-line clinicians and transfusion practitioners. Day 1 - “What Now?” will focus on results and additional data from the trials, to discuss the place of REBOA and cryoprecipitate in future pathways, and also how we can improve our practices for non-compressible haemorrhage, critical bleeding, resuscitation and coagulopathy management in general. Day 2 - “What next?” focuses on challenges in decision-making, and introduces the COMBAT-AID programmes work to bring AI decision support to these patients. Draft programme is available on the registration pagee. The event is funded by the NIHR HTA and the US DoD COMBAT AID programmes. , and will be intentionally interactive – we would really like to hear everyone's views, thoughts and ideas. In-person attendance is limited to just over 150 attendees, and we have reserved spaces for representatives from UK & Ireland MTCs and air ambulance services. Some 40 or so additional places are available. If you would like to join us in person then please indicate whether you would like to be present in person for one or both days. We will inform you by Tuesday 27th February if we have secured space for you. The meeting will also be livestreamed so you can also register for online access. We look forward to seeing you, in person or online, and look forward to two days of intense discussion that will translate into new approaches and guidance for our patients. Please spread the word!
3
74
3
128
46,853
Lots of potential for #MachineLearning in identifying diagnostic #biomarkers, but need appropriate training sets that include confounding comorbidities Our latest: Multi-omic diagnostics of prostate #cancer in the presence of benign prostatic hyperplasia doi.org/10.1016/j.heliyon.20…
1
5
497
Michael Barnes retweeted
🚀Come and join Data Science for Health Equity next week where we will launch our new three-year strategy, laying out our focus areas for 2024–2027. We've also got a special in-conversation sessions from Angela Saini & @Maxi_Macki
🚨ANNOUNCEMENT🚨 DSxHE: Building a world where data improves everyone’s health 🚀 Join us on November 23rd to launch our new strategy and welcome keynote speaker Angela Saini for a discussion on why we categorise patients. eventbrite.co.uk/e/dsxhe-bui…
2
5
813
Michael Barnes retweeted
Great day at #AIDATA23 showcasing our work to a non-scientific audience! Here's our co-lead @thebarneslab w/ @NIHRresearch's Mario Moroso, MELD-B's Simon Fraser, @DECODEproject2's @gcThomasJun, OPTIMAL's @Nirantharakumar & @Monfletch #AI #digitalhealth
4
10
624
Michael Barnes retweeted
Explaining model *predictions* is all well and good – but what about model *uncertainty*? Pleased to announce that our paper on information theoretic Shapley values will be presented at #NeurIPS2023. See you in NOLA! @josh_ke_ohara @NiekTax @ido_guy arxiv.org/abs/2306.05724
4
7
2
28
3,640
I just uploaded a 90 minute tutorial, which is designed to be the one place I point coders at when they ask "hey, tell me everything I need to know about LLMs!" It starts at the basics: the 3-step pre-training / fine-tuning / classifier ULMFiT approach used in all modern LLMs.
73
1,068
66
6,210
958,554
A great opportunity with a great Clinician PI
🌟 Don't miss out on this exciting opportunity for a 5-year post-doctoral fellowship in my group! Perfect for those with expertise in AI, bioinformatics, or robust data science skills. Apply by September 29th. @QMULWHRI #PostDocFellowship #AI #Genomics jobs.ac.uk/job/DBU676/post-d…
2
287
Michael Barnes retweeted
Charting Human Cell Phenoscapes! Can systematic perturbations in human model systems help to define human cell types? Perspective in @ScienceMagazine with @TreutleinLab @josch_f @IHB_Research @ETH_BSSE science.org/doi/full/10.1126… #Organoids #Phenoscaping
5
63
1
234
62,546
It's about 2 years since I declined to renew my IPA licence and tools like metascape and enrichr are making me wonder why it took so long to come to this decision
1
126
Even more fundamentally, I wonder what the bioinformatics community would look like if we paid subscriptions to Ensembl, UCSC, Gtex, Chembl, OpenTargets or maybe even Bioconductor?
94
And have a go with our machine learning notebook, with limited or no coding you can be running an XGBoost model and a SHAP explainer on a transcriptomic dataset in short order. Thanks to @ng88 and @StelMasara for pulling this together
2
139