Application deadline coming up soon for the below machine learning course, with applications to wearables. Very relevant to early career scientists across @HDR_UK ... and also people wishing to conduct analysis of @uk_biobank accelerometer resource

Looking for an open-source, camera-annotated, free-living dataset for wrist-based accelerometry? Check out the @OxWearables Capture-24: Activity tracker dataset for human activity recognition https://t.co/OuDBb92clu via @oxforduni_repo #icampam2022 @ismpb_org

Thank you @GeorgiaTomova, @PWGTennant + @statsmethods for the best week at #LeedsCausalSchool 👩‍💻👩‍🔬🔥 These ideas will be hurting my head for a long time yet 🤯 (but I think today's lesson was that's how it should be? ☑️)

Excited to be on my way to #LeedsCausalSchool 🎉 getting ready to dig some DAGs, learn lots of new things + talk data/epi with new people 😍

Just over 1 day left to apply for this exciting role to lead genomic analysis of wearable phenotypes @bdi_oxford @Oxford_NDPH

There is rightly much talk about using consumer wearable devices for physical activity measurement using various research study designs. We've written a commentary picking up on four key issues we think are relevant for population surveillance https://t.co/1oZOzKAKFf

One week left to apply for this post

Make your day active for health and fun! 2⃣4⃣ strategies to move more and sit less at work, in transport, at home, and in leisure time Free pdf brochure: https://t.co/xJopRHHWl7

Exciting news for all of us interested in physical behaviors and sleep: the launch of the perfect journal to publish our research and reach our readers. ⬇️⬇️⬇️

Hi @RoyalMail, Trying to post a PCR test. Is it really true that not one postbox in Oxford has a collection after 12 on a Saturday or any time on a Sunday?! https://t.co/YeqmHobUoT

📣NEW #OpenAccess Preventing the "24-hour Babel": the need for a consensus on a consistent terminology scheme for physical activity, sedentary behaviour and sleep🏃‍♀️💤 Thanks @DrJenniferCDav1, @LLi_1, @M_Stamatakis👉 https://t.co/6f4uvFKkxY

Great to see this infographic summarizing the contents of our consensus on analytical approaches for accelerometer data! See it online at https://t.co/3ihaYIdxPi @TVoiceOfScience @ortegaporcel

Thanks to @aiden1doherty + Derrick Bennett who supervised this work + to co-authors Shing Chan, @Karl_SmithByrne, @drremar, Mark Woodward, @kazemr + Terry Dwyer. Also to @Oxford_MRC_DTP for funding + @bdi_oxford, @Oxford_NDPH + @uk_biobank Huge thanks to all participants 👏

Thanks to @Oxford_NDPH for this great story on our work:

Data: Free-living acc data with camera-derived ground truth: https://t.co/VnvFlP0rUg Acc data from ~100k participants in @uk_biobank prospective cohort study:

Code: Acc data processing + ML pkg: https://t.co/DLJPAvjiBX UKB data preprocessing pkg: https://t.co/xM1rfrSxqG CoDA analyses pkg: https://t.co/fv3Fc0Odsm Paper repo: https://t.co/sGAgkvV1Fo Paper is out, but code is always work-in-progress- feedback/bug reports welcome!👩‍💻

(2) Reallocating time from other behaviours to moderate-to-vigorous physical activity behaviours was associated with lower CVD risk. Reallocating time from sedentary behaviour to other behaviours was also associated with lower CVD risk.

Key takeaways: (1) Machine-learning behaviour classification achieved good mean accuracy (88%) + Cohen’s kappa (0.80) in free-living wrist accelerometer data against camera ground-truth Confusion matrix ⬇️

In brief, we studied association between device-measured 24-hour movement behaviours and incident CVD, addressing challenges around: (1) classification of behaviours in free-living wrist-worn accelerometer data (2) compositional nature of 24-hour movement behaviour data

Just out in @BJSM_BMJ, our work on device-measured movement behaviours and risk of incident cardiovascular disease (CVD) in 87,498 @uk_biobank participants over >6 years follow-up (4,105 CVD events): https://t.co/SpXk3TxzLz