Gal Sadeh

Sagi Schein, Greg Arutiunian, Vitaly Burshtein et al.

Designing Artificial Intelligence (AI) solutions that can operate in real-world situations is a highly complex task. Deploying such solutions in the medical domain is even more challenging. The promise of using AI to improve patient care and reduce cost has encouraged many companies to undertake such endeavours. For our team, the goal has been to improve early identification of deteriorating patients in the hospital. Identifying patient deterioration in lower acuity wards relies, to a large degree on the attention and intuition of clinicians, rather than on the presence of physiological monitoring devices. In these care areas, an automated tool which could continuously observe patients and notify the clinical staff of suspected deterioration, would be extremely valuable. In order to develop such an AI-enabled tool, a large collection of patient images and audio correlated with corresponding vital signs, past medical history and clinical outcome would be indispensable. To the best of our knowledge, no such public or for-pay data set currently exists. This lack of audio-visual data led to the decision to conduct exactly such study. The main contributions of this paper are, the description of a protocol for audio-visual data collection study, a cloud-architecture for efficiently processing and consuming such data, and the design of a specific data collection device.

Gal Sadeh, Edith Cohen, Haim Kaplan

Influence maximization (IM) is the problem of finding for a given $s\geq 1$ a set $S$ of $|S|=s$ nodes in a network with maximum influence. With stochastic diffusion models, the influence of a set $S$ of seed nodes is defined as the expectation of its reachability over simulations, where each simulation specifies a deterministic reachability function. Two well-studied special cases are the Independent Cascade (IC) and the Linear Threshold (LT) models of Kempe, Kleinberg, and Tardos. The influence function in stochastic diffusion is unbiasedly estimated by averaging reachability values over i.i.d. simulations. We study the IM sample complexity: the number of simulations needed to determine a $(1-ε)$-approximate maximizer with confidence $1-δ$. Our main result is a surprising upper bound of $O( s τε^{-2} \ln \frac{n}δ)$ for a broad class of models that includes IC and LT models and their mixtures, where $n$ is the number of nodes and $τ$ is the number of diffusion steps. Generally $τ\ll n$, so this significantly improves over the generic upper bound of $O(s n ε^{-2} \ln \frac{n}δ)$. Our sample complexity bounds are derived from novel upper bounds on the variance of the reachability that allow for small relative error for influential sets and additive error when influence is small. Moreover, we provide a data-adaptive method that can detect and utilize fewer simulations on models where it suffices. Finally, we provide an efficient greedy design that computes an $(1-1/e-ε)$-approximate maximizer from simulations and applies to any submodular stochastic diffusion model that satisfies the variance bounds.