Magnus Boman

Pathologists routinely perform cancer identification and risk assessments on breast tissue section images and medical patient data. Spatial Transcriptomics (ST) is a novel technology developed in Sweden that allows measuring where in a tissue section a certain RNA was expressed. Our overall aims are to evaluate (a) how the additional consideration of ST gene expression data in addition to tissue section images improves the identification and classification of cancer sub-types in breast tissue sections, (b) if the additional consideration of normal breast tissue adjacent to tumor tissue allows finding cancer onset signatures in the images or in the ST expression data and (c) if considering genome sequence and/or clinical patient data further improves cancer classification.We will computationally analyze our own novel large ST dataset consisting of 48 breast tissue sections from eleven individuals with different cancer types together with non-cancer sections from the same individuals. Whole genome sequencing data is available from the tumor and matched peripheral blood as control from five individuals. Detailed clinical characteristics are available for all individuals.We might be able to give advice on how the novel ST technology can improve traditional cancer identification and classification as well as how the inclusion of additional non-cancer samples, genome sequence information and clinical patient data can improve the classification outcome.

This nursing science precision health project, focus on complex interactions among biological, social, and behavioural factors, and their effects on outcomes. By researching patients’ experience of symptoms and linking these to biological data, we aim to increase the sensitivity and specificity of LC diagnosis and to aid in investigate molecular determinants of common symptoms, e.g. fatigue and cachexia. Furthermore, we aim to develop our measures for use to monitor treatment response in patients receiving immunotherapy. With previous VR funding (2016-1712, 2019-1222), we developed an interactive questionnaire, Peklung, to generate detailed descriptions about early health changes in LC and used it to collect data from ∼700 patients at diagnostic LC work-up

in parallel we also collected plasma samples from  these patients. We apply here for continued VR funding to link patient-reported data from Peklung with biomolecular and imaging profiles, to determine biomarkers of LC and other lung diseases, and to investigate new molecular determinants of symptoms with unclear aetiology and mechanism, e.g. fatigue, cachexia. Establishing symptomatology related to early LC can: decrease diagnostic delay, with increased chance of curative treatment options and shorter time spans to curative or palliative treatment, thus reducing distress for patients/families

connect detailed symptom with omics data to increase understanding of mechanisms of poorly understood symptoms.

Machine learning is already widely used for a variety of tasks. Initial approaches were strictly centralized, but have recently evolved into the so-called federated learning, primarily to address privacy concerns. In this model, multiple clients can collaboratively train a model under the guidance of a centralized server. The clear advantage here is that the data remains with the clients, and they can still influence the global model in their individual training sessions. One of the first deployed systems is at Google, for character recognition on smart phones, but is too slow when operating at massive scale. We have identified a specific societal instance (fatal Tesla Model X crash) in which fast, distributed and federated learning could have prevented a loss of life. Moreover, 21st-century pandemics present dramatic societal problems and require new scalable federated learning techniques. This project aims to develop a highly scalable, flexible, extensible, distributed federated machine learning (Scalable Federated Learning, for short) approach that can directly benefit public health and wellness. Special attention will be paid to the so-called outliers (high-entropy samples). We have two main objectives, and our work is structured around them: 1) Creation of a feasible and flexible scalable model for the life sciences and 2) Development of the scalable, flexible, federated machine learning framework.