Pawel Labaj

The @MCB_UJ is seeking a new director to move it forward in the next step of it’s evolution! Great chance to work with @GlattSebastian @BioPhysMatt @magdamaslon @k_pyrc @RafalMostowy and many more outstanding scientists! tinyurl.com/NewDirectorOfM… @JagiellonskiUni @krakow_pl

Daniel Voskergian (Al-Quds Univ.) presents a Grouping–Scoring–Modeling (GSM) framework at #CAMDA25, #ISBCECCB2025, to predict diabetic complications from synthetic EHRs. Using structured disease-stage groups & XGBoost, models accurately predict key outcomes like CKD & retinopathy

Spiros Denaxas (UCL) opens Day 2 of #CAMDA25, #ISBCECCB2025 with a keynote on the promise & pitfalls of Electronic Health Records in biomedical research. From multidimensional insights to best practices, EHRs are reshaping how we study thousands of conditions simultaneously

Following the tradition, the #CAMDA25 closing ceremony with the awarding of trophies to the best presentations: First proze Anton Pashkov, second Rafael Perez-Estrada, third Owen Vissier. Congratulations!

Yuexi Gu (Xi’an Jiaotong Univ.) presents HI-MGSyn at #CAMDA25, a hypergraph & interaction-aware model for drug synergy prediction. By capturing multi-granular biological signals, it outperforms ML baselines and predicts novel combinations (5 literature validated) #ISMBECCB2025

The #CAMDA25 panel on Health Privacy brought together Spiros Denaxas, Antti Honkela, David Kreil, Wenzhong Xiao & Joaquin Dopazo to debate privacy-preserving ML, synthetic data, and regulatory challenges. A timely conversation on trust, utility & compliance in health AI

Andrew Wicks (DKFZ) explores NMF-based synthetic genomics at #CAMDA25 by integrating differential privacy with zero-inflated sampling, his method offers a practical, privacy-preserving alternative to data sharing, reducing exposure to membership inference risks in genomics.

Steven Golob (Univ. of Washington Tacoma) tackles the data access bottleneck at #CAMD25 by evaluating synthetic data generation (SDG) algorithms for bulk RNA-seq, assessing to what extent they can generate privacy-safe genomics data without sacrificing quality #ISBCECCB2025

Jules Kreuer (Univ. of Tübingen) presents NoisyDiffusion at #CAMDA25 a conditional diffusion model for synthetic gene expression data with built-in differential privacy. High accuracy & low MIA risk show promise for secure, high-utility genomic sharing #ISMBECCB2025

In a second talk, Serghei Mangul reveals gaps in pre-publication omics data sharing: Only 9–23% of datasets are available at preprint time. Early release boosts citations, yet fragmented practices hinder transparency. Time to rethink sharing standards in genomics.

Serghei Mangul (Sage Bionetworks & Univ. of Suceava) analyzes 6M+ papers to map omics data reuse at #CAMDA25. Despite 65% of studies using secondary data, 72% of RNA-seq datasets remain unused Solutions? Better metadata, reuse incentives & formal reusability metrics #ISMBECCB2025

Now at #CAMDA25 #ISMBECCB2025, Doroteya Staykova (Multicore Dynamics) applies Topological Data Analysis to map healthy gut microbiomes. Her methodology reveals two distinct subgroups with unique taxonomic & functional signatures, offering a new perspective on microbiome health.

@ahonkela introduces the #CAMDA25 Health Privacy Challenge, giving a comprehensive discussion of alternative modes of attack, the approach of differential privacy, and the design of an effective Blue Team/Red Team privacy competition. @iscb

Hakime Öztürk (EMBL) introduces the #CAMDA25 Health Privacy Challenge (part of @ELSA_AI) in a Blue vs Red Team setup, participants develop & attack generative models (e.g., VAEs, GANs) for synthetic gene expression data—balancing utility & privacy in biology #ISBCECCB2025

@loucerac introduces the #CAMDA25 Synthetic Electronic Health Records Challenge, discussing how patterns in longitudinal disease trajectories have been distilled from over a million diabetes patients to make them available for public research. @iscb

Antti Honkela (Univ. of Helsinki) kicks off the #CAMDA25 Health Privacy Challenge with a call for responsible, privacy-preserving ML on sensitive health data. He explores how to fairly evaluate privacy–utility trade-offs in trained ML models #ISCBECCB2025

Rafael Pérez Estrada (UNAM) presents an ensemble approach to microbiome health at #CAMDA25, #ISMBECCB2025: By integrating taxonomic (MetaPhlAn) & functional (HUMAnN) data, his Optimized Pathway Ensemble achieves F1 = 0.76. A new web tool compares refined indices across diseases

Kinga Zielińska (Jagiellonian Univ.) opens the #CAMDA25, #ISMBECCB2025 the Gut Microbiota Challenge: Can we build a better microbiome health index? With 4,398 samples & taxonomic + functional data, participants are encouraged to go beyond GMHI & hiPCA

Now at #CAMDA25, #ISMBECCB2025, Khartik Uppalapati (RareGen Youth Network) introduces RDMHI, a rare-disease–specific microbiome health index for PKU, integrating taxonomic, functional & genetic data. RDMHI outperforms GMHI & clinical baselines in forecasting Phe crises.

Vincent Mel (Univ. of Florida) proposes at #CAMDA25 #ISMBECCB2025 a new ensemble-based gut health index by integrating taxonomic & metabolic pathway data. The model outperforms GMHI, hiPCA & Shannon entropy, achieving 72% balanced accuracy and highlighting key microbiome features

Owen Visser (Univ. of Florida) presents an ensemble ML model for AMR prediction at #CAMDA25, #ISMBECCB2025. Trained on strain-specific markers & AMR gene classes, achieving up to 98.2% accuracy (A. baumannii). Permutation analysis reveals key resistance genes in diverse pathogens