Machine learning and artificial intelligence in the service of medicine: Necessity or potentiality?

MOTIVATION: As a result of the worldwide health care system digitalization trend, the produced healthcare data is estimated to reach as much as 2314 Exabytes of new data generated in 2020. The ongoing development of intelligent systems aims to provide better reasoning and to more efficiently use the data collected. This use is not restricted retrospective interpretation, that is, to provide diagnostic conclusions. It can also be extended to prospective interpretation providing early prognosis. That said, physicians who could be assisted by these systems find themselves standing in the gap between clinical case and deep technical reviews. What they lack is a clear starting point from which to approach the world of machine learning in medicine. METHODOLOGY AND MAIN STRUCTURE: This article aims at providing interested physicians with an easy-to-follow insight of Artificial Intelligence (AI) and Machine Learning (ML) use in the medical field, primarily over the last few years. To this end, we first discuss the general developmental paths concerning AI and ML concept usage in healthcare systems. We then list fields where these technologies are already being put to the test or even applied such as in Hematology, Neurology, Cardiology, Oncology, Radiology, Ophthalmology, Cell Biology and Cell Therapy.

Learning, planning, and control in a monolithic neural event inference architecture.

We introduce REPRISE, a REtrospective and PRospective Inference SchEme, which learns temporal event-predictive models of dynamical systems. REPRISE infers the unobservable contextual event state and accompanying temporal predictive models that best explain the recently encountered sensorimotor experiences retrospectively. Meanwhile, it optimizes upcoming motor activities prospectively in a goal-directed manner. Here, REPRISE is implemented by a recurrent neural network (RNN), which learns temporal forward models of the sensorimotor contingencies generated by different simulated dynamic vehicles. The RNN is augmented with contextual neurons, which enable the encoding of distinct, but related, sensorimotor dynamics as compact event codes. We show that REPRISE concurrently learns to separate and approximate the encountered sensorimotor dynamics: it analyzes sensorimotor error signals adapting both internal contextual neural activities and connection weight values. Moreover, we show that REPRISE can exploit the learned model to induce goal-directed, model-predictive control, that is, approximate active inference: Given a goal state, the system imagines a motor command sequence optimizing it with the prospective objective to minimize the distance to the goal. The RNN activities thus continuously imagine the upcoming future and reflect on the recent past, optimizing the predictive model, the hidden neural state activities, and the upcoming motor activities. As a result, event-predictive neural encodings develop, which allow the invocation of highly effective and adaptive goal-directed sensorimotor control.

Adding phosphorylation events to the core oscillator driving the cell cycle of fission yeast.

Much is known about the regulatory elements controlling the cell cycle in fission yeast (Schizosaccharomyces pombe). This regulation is mainly done by the (cyclin-dependent kinase/cyclin) complex (Cdc2/Cdc13) that activates specific target genes and proteins via phosphorylation events during the cell cycle in a time-dependent manner. However, more work is still needed to complement the existing gaps in the current fission yeast gene regulatory network to be able to overcome abnormalities in its growth, repair and development, i.e. explain many phenomena including mitotic catastrophe. In this work we complement the previously presented core oscillator of the cell cycle of fission yeast by selected phosphorylation events and study their effects on the temporal evolution of the core oscillator based Boolean network. Thereby, we attempt to establish a regulatory link between the autonomous cell cycle oscillator and the remainder of the cell. We suggest the unclear yet regulatory effect of phosphorylation on the added components, and discuss many unreported points regarding the temporal evolution of the cell cycle and its components. To better visualize the results regardless of the programming background we developed an Android application that can be used to run the core and extended model of the fission yeast cell cycle step by step.