From the Neuronauts blog
Notes on clinical AI, decision support and the discipline of shipping trustworthy healthcare software.
Learning More From Less: The Rise of Active Learning
Ali Güneş on how Active Learning gets similar or better performance from far less labeled data — the query strategies, the deep learning variants, and two case studies from his own published work.
The Age of Augmented Expertise in Bioinformatics
Computational biologist Orhan Nedim Kurt on what AI actually speeds up in bioinformatics — and what still belongs entirely to the expert.
Human-in-the-loop: why clinical AI should know when to stay quiet
The measure of clinical AI is not only how often it is right, but whether it knows when it is unsure, defers to the clinician, and leaves a trail.
Institution-specific clinical apps: built around the workflow, not the other way round
Generic software asks the clinic to change how it works. The better model is the reverse — apps shaped to the ward, the round and the checklist that already exist.
Turning scattered clinical data into analysis-ready datasets — without breaking KVKK
Most institutions do not lack data — they lack data they can trust and legally use. The work is in cleaning, structuring and de-identifying it by design.
HBYS integration: the unglamorous work that decides whether your AI ships
Clinical AI lives or dies on its connection to the hospital information system. A field guide to HL7, FHIR and the realities of HBYS in Turkish hospitals.
PACS in the AI era: getting models into the radiology workflow
Most radiology AI does not fail on accuracy. It fails on the last mile — the path between the model and the radiologist's PACS worklist.
AI in bioinformatics: from variant calling to clinical interpretation
Sequencing stopped being the bottleneck years ago. Interpretation is the bottleneck — and that is where machine learning is quietly changing genomic medicine.
Agentic automation for clinic operations
Agent-based AI can absorb the repetitive administrative load around care — if you scope it tightly and keep a human checkpoint.
What makes a clinical decision support system trustworthy
Accuracy is table stakes. Adoption depends on calibration, transparency, and fitting the workflow clinicians already have.
How AI is reshaping diagnostic pathology
A practical look at where decision support adds real value in the lab — and where the human pathologist stays firmly in the loop.