Neural Bayes Estimators (NBEs) provide a simple simulation-based approach to phylodynamic inference.
For comparisons with gold‑standard MCMC methods, speed, and additional details on the simulation studies you'll need to check out the paper 🙂
Feedback is very welcome!
Simulation studies show the method has low bias and has well-calibrated 95% CIs.
We also evaluated performance under model misspecification.
Performance degraded, but fine‑tuning was fast and restored accuracy, suggesting a useful embedding was learned.
To process phylogenetic trees we use a recursive neural network architecture.
This allows the model to ingest the tree structure directly rather than relying on hand-crafted summary statistics.
Trees contain information about epidemic dynamics, but extracting this can be challenging.
We trained a neural Bayes estimator to predict several epidemic quantities:
- time-varying reproduction number (R_t)
- prevalence of infection
- cumulative number of infections
Excited to share a new preprint with the amazing Thomas Williams (from @macsys.bsky.social ) and Louis du Plessis (from @bsse.ethz.ch ) on neural inference for phylodynamics. 🧬🌳
arXiv: arxiv.org/abs/2603.08345
Promotional graphic advertising dates for MASAMB 2026 7 - 8 September 2026 at Sidney Sussex College, Cambridge More info coming soon - www.mrc-bsu.cam.ac.uk
Exciting news! ✨ We will be hosting MASAMB 2026.
Save the dates: 7 - 8 September, Sidney Sussex College, Cambridge
More info, including how to submit an abstract, will be coming soon 👇
www.mrc-bsu.cam.ac.uk/events
#MASAMB2026
PhD position (Manchester, UK)
Modelling reactive response in outbreaks of respiratory infections in vulnerable settings
with Lorenzo Pellis, Ian Hall, Thomas House
at University of Manchester
More details: http://iddjobs.org/jobs/2462
Excited to start my @ukri.org Future Leaders Fellowship at @kingscollegelondon.bsky.social and lead the GeoAI for Health Equity Lab! We are hiring for 3 posts!! Join me to advance GeoAI research that strengthens epidemic preparedness and reduces healths inequities. 1/n
Happy #WorldStatisticsDay2025! 🎉
At the BSU, you can find a fantastic group of passionate people driven to improve the way we look at scientific problems and design methods to help us understand the deluge of data we are surrounded by.
Thanks again for the invitation and the lively discussion
Sounds like a fun conference, bet there are a ton of great speakers
@aezarebski.bsky.social will give 2 talks in 🇨🇭:
🔹 Bern – 31/3 15:15
🔹 Basel – 16/4, 11:15
💡 Talk: Narrow AI for Phylodynamics
This visit is supported by SIB’s Speaker Roadshow Initiative, which coordinates multiple talks at partner institutions and promotes sustainable travel.
A huge thanks to Louis du Plessis (ETH) and Claudia Bank's lab (UoBern) for this awesome invitation, and the @sib.swiss for their support via the Roadshow Initiative. (Which is really an great idea!)
We are advertising a PhD position! The project will develop mathematical models for influenza infection within the host, incorporating the effects of antiviral treatment. Domestic and international applicants welcome.
Lots of studies in epidemiology rely on people being able to assess age of others - please participate in this survey to help get better error models for such data!
warwick.co1.qualtrics.com/jfe/form/SV_...
Side effect of report marking?
Yep, couldn't resist responding with the reverse. Both are interesting reads though.
This one is interesting too:
Estimating the duration of RT-PCR positivity for SARS-CoV-2 from doubly interval censored data with undetected infections
doi.org/10.48550/arX...
And a bonus post to share links to my coauthors (because apparently I can't use bluesky search properly...)
@nefeltellioglu.bsky.social who kept this project alive and kicking, and @camzachreson.bsky.social who masterminded it and you should follow if you're interested in this topic.
While not exhaustive, our results suggest (for COVID-19) outcome equity can co-exist with utilitarian minimization of disease burden. We hope to expand our approach to other public health values (e.g. autonomy, privacy) and other interventions.
I'd love to hear your thoughts! 💬
In the aftermath of the COVID-19 pandemic, infectious disease modelling is taking a hard look at health equity. We did this work to help understand the potential impacts of considering equity.
Exploring different weightings on equity and aggregate burden in a model of COVID-19 in Melbourne. We found a trade-off in equity of vaccine-induced burden, and total burden. Importantly, there was no trade-off between total burden and equity of disease burden 💉
We considered: total days spent in hospital due to infection, and rare vaccine side effects; how fairly the infection burden shared; and how fairly vaccination burden is shared. The triangle shows possible weightings, much of the literature focuses on aggregate burden.🔬
We look at optimizing public health interventions while prioritizing both equity and clinical burden. Basically “what level of vaccination should you aim for in the different age groups if they have different risk profiles?”🤒😷
A huge thank you to Nefel Tellioglu, Jessica Stockdale, Julie Spencer, Wasiur Rahman Khuda Bukhsh, @joelcmiller.bsky.social and Cameron Zachreson for including me on this @matrix-inst.bsky.social project we just preprinted results from
t.co/9ioaiNWMQa 🤓
[Please share] We're hiring a postdoc in Mathematical Biology at The University of Melbourne! Come and work with us on a collaborative project looking at how to predict biological behaviour across multiple scales. Feel free to reach out, details are here: jobs.unimelb.edu.au/en/job/91913...
@gerrythill.bsky.social you're such a trend setter!
