Language Models Are An Effective Patient Representation Learning Technique For Electronic Health Record Data Ethan Steinberg, Ken Jung, Jason A. Fries, Conor K. Corbin, Stephen R. Pfohl, Nigam H. Shah January 16, 2020 Abstract Widespread adoption of electronic health records (EHRs) has fueled development of clinical outcome models using machine learning. However, patient EHR data are complex, and how to optimally represent them is an open question. This complexity, along with often small training set sizes available to train these clinical outcome models, are two core challenges for training high quality models. In this paper, we demonstrate that learning generic representations from the data of all the patients in the EHR enables better performing prediction models for clinical outcomes, allowing for these challenges to be overcome. We adapt common representation learning techniques used in other domains and find that representations inspired by language models enable a 3.5% mean improvement in AUROC on five clinical outcomes compared to standard baselines, with the average improvement rising to 19% when only a small number of patients are available for training a prediction model for a given clinical outcome. 1 Introduction The widespread adoption of electronic health records (EHRs) has created opportunities for using machine learning to reduce healthcare costs and improve quality of care. EHR data have been used to learn prediction models for clinical outcomes such as mortality [1], sepsis [2], 30-day readmission [3] and others [4, 5]. However, the complexity of patient data poses many obstacles to its effective use. Patient records in EHRs are variable length, high dimensional and sparse, with complex temporal and hierarchical structure. They are comprised of irregularly spaced visits spread across years, with each visit consisting of a subset of thousands of possible diagnosis, procedure, and medication codes as well as lab values and unstructured data such as text or images. In contrast, most off-the-shelf machine learning algorithms expect a fixed length vector of features as input. Manually defining a transformation of patient records into such a representation beyond simple binned counts is time consuming and outcome-dependent, leaving much of the temporal and hierarchical structure of EHRs underutilized when building machine learning models. The challenge of representing EHR data can be addressed by using neural networks to automatically learn how to featurize patient data while learning a model for a given clinical outcome (e.g., mortality or 30 day readmissions) [4].

Federated learning is an emerging research paradigm to train models on private data distributed over multiple devices. A key challenge involves keeping private all the data on each device and training a global model only by communicating parameters and updates. Overcoming this problem relies on the global model being sufficiently compact so that the parameters can be efficiently sent over communication channels such as wireless internet. Given the recent trend towards building deeper and larger neural networks, deploying such models in federated settings on real-world tasks is becoming increasingly difficult. To this end, we propose to augment federated learning with local representation learning on each device to learn useful and compact features from raw data. As a result, the global model can be smaller since it only operates on higher-level local representations. We show that our proposed method achieves superior or competitive results when compared to traditional federated approaches on a suite of publicly available real-world datasets spanning image recognition (MNIST, CIFAR) and multimodal learning (VQA). Our choice of local representation learning also reduces the number of parameters and updates that need to be communicated to and from the global model, thereby reducing the bottleneck in terms of communication cost. Finally, we show that our local models provide flexibility in dealing with online heterogeneous data and can be easily modified to learn fair representations that obfuscate protected attributes such as race, age, and gender, a feature crucial to preserving the privacy of on-device data.

With the success of deep neural networks, Neural Architecture Search (NAS) as a way of automatic model design has attracted wide attention. As training every child model from scratch is very time-consuming, recent works leverage weight-sharing to speed up the model evaluation procedure. These approaches greatly reduce computation by maintaining a single copy of weights on the super-net and share the weights among every child model. However, weight-sharing has no theoretical guarantee and its impact has not been well studied before. In this paper, we conduct comprehensive experiments to reveal the impact of weight-sharing: (1) The best-performing models from different runs or even from consecutive epochs within the same run have significant variance; (2) Even with high variance, we can extract valuable information from training the super-net with shared weights; (3) The interference between child models is a main factor that induces high variance; (4) Properly reducing the degree of weight sharing could effectively reduce variance and improve performance.

Its impact is drastic and real: Youtube's AIdriven recommendation system would present sports videos for days if one happens to watch a live baseball game on the platform [1]; email writing becomes much faster with machine learning (ML) based auto-completion [2]; many businesses have adopted natural language processing based chatbots as part of their customer services [3]. AI has also greatly advanced human capabilities in complex decision-making processes ranging from determining how to allocate security resources to protect airports [4] to games such as poker [5] and Go [6]. All such tangible and stunning progress suggests that an "AI summer" is happening. As some put it, "AI is the new electricity" [7]. Meanwhile, in the past decade, an emerging theme in the AI research community is the so-called "AI for social good" (AI4SG): researchers aim at developing AI methods and tools to address problems at the societal level and improve the wellbeing of the society.

One of the main reasons why RL has worked so well in these applications is that we are able simulate millions of interactions with the environment in a relatively short period of time, allowing the agent to experience a large number of different situations in the environment and learn the consequences of its actions. In many real world applications, however, where the agent interacts with the physical world, it might not be easy to generate such a large number of interactions. The time and cost associated with training such systems could render RL an unfeasible approach for training in large scale. As a concrete example, consider training a large number of humanoid robots (agents) to move quickly, as in the Robocup competition [ Farchy et al., 2013 ] . Although the agents have similar dynamics, subtle variations mean that a single policy shared across all agents would not be an effective solution.

Cameras with artificial intelligence (AI) software that the South Korean government claims can detect the likelihood of crime will be installed in Seoul within the year. The Seocho District of South Korea's capital and Electronics and Telecommunications Research Institute (ERTI), a national research institute, said they will install 3,000 cameras at the district by July. The cameras will use AI software that processes the location, time, and behaviour patterns of passersby to measure the likelihood of a crime taking place. The cameras will automatically measure whether somebody is walking normally or tailing someone. It will also detect what passersby are wearing -- such as hats, masks, or glasses -- and what they are carrying with them such as bags or dangerous objects that have a strong possibility of being used to commit a crime.

Music and dance are so deeply embedded in the human experience that we almost take them for granted. They're distinct from one another, but intimately related: Music -- arrangements of sound over time -- causes us to move our bodies in space. Without knowing it, we track pulse, tempo and rhythm, and we move in response. But only recently have scientists developed the tools, and the inclination, to quantitatively study the human response to music in its many forms. It's a research program that relies on a wide array of approaches, employing techniques from the study of perception and cognition to those of neurobiology and neuroimaging, with additional insights from psychophysics, evolutionary psychology and animal studies.

Sundar Pichai, the chief executive of Google, has been put in charge of its parent company Alphabet, after co-founders Larry Page and Sergey Brin announced they were stepping down. The 47-year-old said the pair had set up a "strong foundation" on which he would "continue to build". Pichai's life story is remarkable, and his rise to the top of Google is an endorsement of India's standing in the global technology industry - and equally, a reassuring reminder of the so-called "American Dream". Pichai was born and schooled in Chennai, India. He captained his school's cricket team, leading it to win regional competitions.

Sundar Pichai, the chief executive of Google, has been put in charge of its parent company Alphabet, after co-founders Larry Page and Sergey Brin announced they were stepping down. The 47-year-old said the pair had set up a "strong foundation" on which he would "continue to build". Pichai's life story is remarkable, and his rise to the top of Google is an endorsement of India's standing in the global technology industry - and equally, a reassuring reminder of the so-called "American Dream". Pichai was born and schooled in Chennai, India. He captained his school's cricket team, leading it to win regional competitions.

Cyclists, whether they be food delivery riders or MAMILs, are infamous for doing everything they can to conserve their hard-won speed, even if it means running a red light or careering into the way of pedestrians on the footpath. But new work from IBM Research Australia and RMIT's Exertion Games Lab, however, is looking to avoid tiresome stops or dangerous behaviour by using artificial intelligence (AI) to catch the'green wave' of traffic signals. It's well known many cyclists jump traffic signals or make legally questionable deviations to maintain momentum getting from A to B. If you're an underpaid international student under Dickensian food delivery conditions, there's simply no other way. That could be about to change. In a project dubbed'Ari the e-bike,' the researchers used traffic data and'green wave' modelling from VicRoads and internet of things (IoT) technologies to help the rider regulate their speed to match cycles of green traffic lights.