Search

Biological computing - Wikipedia

The document describes biocomputers that use biological materials to perform computations, defining nanobiotechnology, and outlining three main biocomputer types—biochemical, biomechanical, and bioelectronic—along with network-based biocomputation and how inputs yield outputs via chemical, mechanical, or electrical signals.

They put living human neurons into a computer: their goal could transform AI

Translated title. Original: Ils ont mis des neurones humains vivants dans un ordinateur : leur objectif pourrait transformer l'IA

Living human neurons on silicon chips form 'wetware' to run AI, offering energy efficiency and new capabilities, though current performance trails conventional GPUs; potential uses span medicine, neuroscience, robotics, and cybersecurity.

Hugging Face presents Carbon-X, a system… | Hacks.gr

Hugging Face announces Carbon-X, a locally-run AI system for DNA-based conversational queries, trained on 300B data chunks and designed to read six-letter DNA blocks; virus data was excluded. While intended to assist drug and therapy design, its laboratory validity remains unverified, and the system runs offline with 14B parameters.

Computational Neuroscience: Models, Applications, and the Future

Translated title. Original: Neurociencia Computacional: Modelos, Aplicaciones y Futuro

Computational neuroscience combines biology, math, and computing to build and test biologically realistic brain models, using EEG/MEG data and simulations to understand neural processes and inform medical and educational tools, while addressing ethical issues and pursuing advanced computing to scale models.

MMK researchers appointed DDLS Fellow – receive long-term research funding | Karolinska Institutet News

Translated title. Original: MMK-forskare utsedd till DDLS Fellow – får långsiktigt forskningsstöd | Karolinska Institutet Nyheter

Karolinska Institutet announces that David Marlevi and the translational cardiovascular imaging team have been named DDLS Fellows, receiving long-term funding to advance data-driven life sciences. The program supports career development and multi-year research funding to integrate biology, big data, computation, and AI for improved health outcomes.

Anthropic's Claude artificial intelligence model makes a first in biology by mapping the 3D structure of a prokaryotic capsule particle - Artificial Intelligence News | airehber.com.tr

Translated title. Original: Anthropic’in Claude yapay zeka modeli biyolojide bir ilke imza atarak prokaryotik kasa parçacığının 3D yapısını haritalandırdı - Yapay Zeka Haberleri | airehber.com.tr

Anthropic reports using the Claude model to map the first 3D structure of a prokaryotic vault particle. The work speeds up purification from days to 90 minutes and reduces modeling time from weeks to 34 hours by combining protein database search, DNA design, purification optimization, and cryo-EM data processing.

Google DeepMind Presents AlphaGenome 2 | IA en un minuto

Google DeepMind says AlphaGenome 2 predicts how DNA variants affect specific cell types, with improved variant detection in its tests. The model is available by API for non-commercial research and in a private preview for commercial research; it is not approved for clinical use.

Microsoft AI Hits 83% on Past Cases, Science Paper | TSN

Microsoft's AI diagnostic system, DxO, scored higher than generalist physicians on benchmark medical cases, using a GPT-5-assisted evaluation. While promising, the study notes limits and plans a prospective clinical validation with Beth Israel Deaconess.

Alphabet's DeepMind unveils new AlphaGenome 2 model aimed at reading DNA cells (GOOG:NASDAQ) | Seeking Alpha

DeepMind unveiled AlphaGenome 2, an AI model aimed at reading DNA in cells. The announcement describes a shift from broad tissue averages to individual cell types and states.

New AI model detects hidden diseases in routine CT scans

A Penn Medicine study introduces Percival, an AI model that analyzes routine CT scans to uncover hidden health risks by learning from imaging data and linking it to labs and outcomes. Trained on hundreds of thousands of scans from the Penn Medicine BioBank, Percival aims to enable opportunistic screening and improve early detection, with open-source code and ongoing clinical testing.