Search

Artificial intelligence (AI) is one of the key technological innovations shaping our modern world. The growing availability of digital health data, combined with the newest technical advancements in computing and data science, form the perfect environment for data-driven, AI-powered medicine to flourish. Many of the challenges presented by AI arise from its very foundation, from biases to transparency or generalizability. This chapter therefore focuses on presenting the basic ideas behind AI algorithms. In particular, it focuses on the different types of learning (supervised vs. unsupervised, shallow vs. deep), and discusses some of the most popular ways in which AI methods have been applied in healthcare. We then discuss the main technological pitfalls, and summarize them into four key questions on transparency, overfitting, performance and robustness that AI models with translational ambitions should be able to address adequately. Finally, AI has the potential to go beyond simple applications, and become one of the key tools to connect medical research and clinical practice. We suggest that, if the field progresses steadily despite all the complexities, it could become the backbone of a knowledge-generating feedback loop between the two.

Ancient genomes have revolutionized our understanding of Holocene prehistory and, particularly, the Neolithic transition in western Eurasia. In contrast, East Asia has so far received little attention, despite representing a core region at which the Neolithic transition took place independently similar to 3 millennia after its onset in the Near East. We report genome-wide data from two hunter-gatherers from Devil's Gate, an early Neolithic cave site (dated to similar to 7.7 thousand years ago) located in East Asia, on the border between Russia and Korea. Both of these individuals are genetically most similar to geographically close modern populations from the Amur Basin, all speaking Tungusic languages, and, in particular, to the Ulchi. The similarity to nearby modern populations and the low levels of additional genetic material in the Ulchi imply a high level of genetic continuity in this region during the Holocene, a pattern that markedly contrasts with that reported for Europe., Postprints der Universität Potsdam : Mathematisch-Naturwissenschaftliche Reihe, 791

The agricultural transition profoundly changed human societies. We sequenced and analysed the first genome (1.39x) of an early Neolithic woman from Ganj Dareh, in the Zagros Mountains of Iran, a site with early evidence for an economy based on goat herding, ca. 10,000 BP. We show that Western Iran was inhabited by a population genetically most similar to hunter-gatherers from the Caucasus, but distinct from the Neolithic Anatolian people who later brought food production into Europe. The inhabitants of Ganj Dareh made little direct genetic contribution to modern European populations, suggesting those of the Central Zagros were somewhat isolated from other populations of the Fertile Crescent. Runs of homozygosity are of a similar length to those from Neolithic farmers, and shorter than those of Caucasus and Western Hunter-Gatherers, suggesting that the inhabitants of Ganj Dareh did not undergo the large population bottleneck suffered by their northern neighbours. While some degree of cultural diffusion between Anatolia, Western Iran and other neighbouring regions is possible, the genetic dissimilarity between early Anatolian farmers and the inhabitants of Ganj Dareh supports a model in which Neolithic societies in these areas were distinct., Postprints der Universität Potsdam : Mathematisch-Naturwissenschaftliche Reihe, 952