In hematology, computational models offer significant potential to improve diagnostic accuracy, streamline workflows, and reduce the tedious work of analyzing single cells in peripheral blood or bone marrow smears. However, clinical adoption of computational models has been hampered by the lack of generalization due to large batch effects, small dataset sizes, and poor performance in transfer learning from natural images. To address these challenges, we introduce DinoBloom, the first foundation model for single cell images in hematology, utilizing a tailored DINOv2 pipeline. Our model is built upon an extensive collection of 13 diverse, publicly available datasets of peripheral blood and bone marrow smears, the most substantial open-source cohort in hematology so far, comprising over 380,000 white blood cell images. To assess its generalization capability, we evaluate it on an external dataset with a challenging domain shift. We show that our model outperforms existing medical and non-medical vision models in (i) linear probing and k-nearest neighbor evaluations for cell-type classification on blood and bone marrow smears and (ii) weakly supervised multiple instance learning for acute myeloid leukemia subtyping by a large margin. A family of four DinoBloom models (small, base, large, and giant) can be adapted for a wide range of downstream applications, be a strong baseline for classification problems, and facilitate the assessment of batch effects in new datasets. All models are available at github.com/marrlab/DinoBloom.

"Together, I think technology and AI will transform the way we diagnose and we treat in the coming years. And this is really enabling clinical care to become very tailor-made, individualized. But even more than that, it enables the patient to get access to the information and the doctors to make better decisions." Intro: Welcome to the Agile Digital Transformation podcast, where we explore different aspects of digital transformation and digital experience with your host, Tim Butara, content and community manager at Agiledrop. Thank you for tuning in.

Artificial intelligence (AI) is about to make itself indispensable in the health care sector. Examples of successful applications or promising approaches range from the application of pattern recognition software to pre-process and analyze digital medical images, to deep learning algorithms for subtype or disease classification, and digital twin technology and in silico clinical trials. Moreover, machine-learning techniques are used to identify patterns and anomalies in electronic health records and to perform ad-hoc evaluations of gathered data from wearable health tracking devices for deep longitudinal phenotyping. In the last years, substantial progress has been made in automated image classification, reaching even superhuman level in some instances. Despite the increasing awareness of the importance of the genetic context, the diagnosis in hematology is still mainly based on the evaluation of the phenotype. Either by the analysis of microscopic images of cells in cytomorphology or by the analysis of cell populations in bidimensional plots obtained by flow cytometry. Here, AI algorithms not only spot details that might escape the human eye, but might also identify entirely new ways of interpreting these images. With the introduction of high-throughput next-generation sequencing in molecular genetics, the amount of available information is increasing exponentially, priming the field for the application of machine learning approaches. The goal of all the approaches is to allow personalized and informed interventions, to enhance treatment success, to improve the timeliness and accuracy of diagnoses, and to minimize technically induced misclassifications. The potential of AI-based applications is virtually endless but where do we stand in hematology and how far can we go?

As a subfield of machine learning, \emph{reinforcement learning} (RL) aims at empowering one's capabilities in behavioural decision making by using interaction experience with the world and an evaluative feedback. Unlike traditional supervised learning methods that usually rely on one-shot, exhaustive and supervised reward signals, RL tackles with sequential decision making problems with sampled, evaluative and delayed feedback simultaneously. Such distinctive features make RL technique a suitable candidate for developing powerful solutions in a variety of healthcare domains, where diagnosing decisions or treatment regimes are usually characterized by a prolonged and sequential procedure. This survey will discuss the broad applications of RL techniques in healthcare domains, in order to provide the research community with systematic understanding of theoretical foundations, enabling methods and techniques, existing challenges, and new insights of this emerging paradigm. By first briefly examining theoretical foundations and key techniques in RL research from efficient and representational directions, we then provide an overview of RL applications in a variety of healthcare domains, ranging from dynamic treatment regimes in chronic diseases and critical care, automated medical diagnosis from both unstructured and structured clinical data, as well as many other control or scheduling domains that have infiltrated many aspects of a healthcare system. Finally, we summarize the challenges and open issues in current research, and point out some potential solutions and directions for future research.

Journal of Parallel and Distributed Computing 104, 19-35.

While generative models such as Latent Dirichlet Allocation (LDA) have proven fruitful in topic modeling, they often require detailed assumptions and careful specification of hyperparameters. Such model complexity issues only compound when trying to generalize generative models to incorporate human input. We introduce Correlation Explanation (CorEx), an alternative approach to topic modeling that does not assume an underlying generative model, and instead learns maximally informative topics through an information-theoretic framework. This framework naturally generalizes to hierarchical and semi-supervised extensions with no additional modeling assumptions. In particular, word-level domain knowledge can be flexibly incorporated within CorEx through anchor words, allowing topic separability and representation to be promoted with minimal human intervention. Across a variety of datasets, metrics, and experiments, we demonstrate that CorEx produces topics that are comparable in quality to those produced by unsupervised and semi-supervised variants of LDA.

Researchers have used machine learning to identify how individual stroke patients might respond to different medications, based on the unique structure of their brain. An experiment by University College London (UCL) found that applying computer intelligence to data from from people who had suffered a stroke allowed researchers to see what effect drugs had on brains with varying patterns of damage. For the study, a machine learning algorithm was applied to CT and MRI scans of 1172 stroke patients and mapped the anatomical pattern of damage throughout the brain of each individual. The researchers then simulated the effects of certain hypothetical drugs, to see if any reactions that would have been missed by conventional methods could be identified. They found that the algorithm was particularly advantageous when looking at medication effects that reduced the size of lesions in patients' brains.