Deep learning algorithm could aid drug development Stanford News


Artificially intelligent algorithms can learn to identify amazingly subtle information, enabling them to distinguish between people in photos or to screen medical images as well as a doctor. But in most cases their ability to perform such feats relies on training that involves thousands to trillions of data points. This means artificial intelligence doesn't work all that well in situations where there is very little data, such as drug development. Vijay Pande, professor of chemistry at Stanford University, and his students thought that a fairly new kind of deep learning, called one-shot learning, that requires only a small number of data points might be a solution to that low-data problem. Stanford chemistry Professor Vijay Pande and his students see a future for machine learning in the early stages of drug development.

How AI can help to optimise processes


As new and tailored variants of medicines are developed can traditional clinical trial management solutions still meet the evolving needs of life sciences companies? However, what if different types of content and workflow management solutions could dynamically alter flows based on its own perceptible AI? However, there is an opportunity with AI for process flows to dynamically target the production of different types of flu vaccine automatically. Today, Enterprise Content Management (ECM) processes still require someone with intelligence to understand the defined flow (for example managing clinical trials), as it is a highly regulatory process that has a lot of governance and control.

Temporal Vaccination Games under Resource Constraints

AAAI Conferences

The decision to take vaccinations and other protective interventions for avoiding an infection is a natural game-theoretic setting. Most of the work on vaccination games has focused on decisions at the start of an epidemic. However, a lot of people defer their vaccination decisions, in practice. For example, in the case of the seasonal flu, vaccination rates gradually increase, as the epidemic rate increases. This motivates the study of temporal vaccination games, in which vaccination decisions can be made more than once. An important issue in the context of temporal decisions is that of resource limitations, which may arise due to production and distribution constraints. While there has been some work on temporal vaccination games, resource constraints have not been considered. In this paper, we study temporal vaccination games for epidemics in the SI (susceptible-infectious) model, with resource constraints in the form of a repeated game in complex social networks, with budgets on the number of vaccines that can be taken at any time. We find that the resource constraints and the vaccination and infection costs have a significant impact on the structure of Nash equilibria (NE). In general, the budget constraints can cause NE to become very inefficient, and finding efficient NE as well as the social optimum are NP-hard problems. We develop algorithms for finding NE and approximating the social optimum. We evaluate our results using simulations on different kinds of networks.

'Artificial life' breakthrough announced by scientists - BBC News


The researchers constructed a bacterium's "genetic software" and transplanted it into a host cell. But the researchers hope eventually to design bacterial cells that will produce medicines and fuels and even absorb greenhouse gases. Dr Venter told BBC News: "We've now been able to take our synthetic chromosome and transplant it into a recipient cell - a different organism. Dr Helen Wallace from Genewatch UK, an organisation that monitors developments in genetic technologies, told BBC News that synthetic bacteria could be dangerous.