Argumentation provides a powerful mechanism for dealing with incomplete, possibly inconsistent information and for the resolution of conflicts and differences of opinion amongst different parties. Further, it is useful for justifying outcomes. Thus, argumentation can support several aspects of decision-making, either by individual entities performing critical thinking (needing to evaluate pros and cons of conflicting decisions) or by multiple entities dialectically engaged to come to mutually agreeable decisions (needing to assess the validity of information the entities become aware of and resolve conflicts), especially when decisions need to be transparently justified (e.g. in medicine). Because of its potential to support decision-making when transparently justifying decisions is essential, the use of argumentation has been considered in a number of settings, including medicine, law, e-procurement, e-business and design rationale in engineering. Potential users of existing argumentation-based decision-making methods are empowered by transparent methods, afforded by argumentation, but lack either means of formal evaluation sanctioning decisions as (individually or collectively) rational or a computational framework for supporting automation. The combination of these three features (transparency, rationality and computational tools for automation) is essential for argumentation-based decision-making to have a fruitful impact on applications. Indeed, for example, a medical practitioner would not find a "black-box" recommended decision useful, but he/she would also not trust a fully transparent, dialectically justified decision unless he/she were sure that this is the best one (rational). In addition, the plethora of information doctors need to take into account nowadays to make decisions requires automated support. TRaDAr aims at providing methods and prototype systems for various kinds of argumentation-based (individual and collaborative) decision-making that generate automatically transparent, rational decisions, while developing case studies in smart electricity and e-health to inform and validate methods and systems. In this context, TRaDAr's technical objectives are: (O1) to provide novel argumentation-based formulations of decision problems for individual and collaborative decision-making; (O2) to study formal properties of the formulations at (O1), sanctioning the rationality of decisions; (O3) to provide real-world case studies in smart electricity and e-health for (individual and collaborative) decision-making, using the formulations at (O1) and demonstrating the importance of the properties at (O2) as well as the transparent nature of argumentation-based decision-making; (O4) to define provably correct algorithms for the formulations at (O1), supporting rational and transparent (individual and collaborative) decision-making; (O5) to implement prototype systems incorporating the computational methods at (O4), and use these systems to demonstrate the methodology at (O1-O2) for the case studies at (O3). The project intends to develop novel techniques within an existing framework of computational argumentation, termed assumption-based argumentation, towards the achievements of these objectives, and adapting notions and techniques from classical (quantitative) decision theory and mechanism design in economics. The envisaged TRaDAr's methodology and systems will contribute to a sustainable society supported by the digital economy, and in particular they will support people in making informed choices. The project will focus on demonstrating the proposed techniques in specific case studies (smart electricity and e-health for breast cancer) in two chosen application areas (digital economy and e-health), but its outcomes could be far-reaching into other case studies (e.g. in other areas of medicine) as well as other sectors (e.g. in engineering, for supporting decisions on design choices).

In this fellowship, I will use a radical new approach to improve the radiotherapy treatment of patients suffering from inoperable non-small cell lung cancer (NSCLC). NSCLC is a cancer of unmet need for which the actual chemo-radiotherapy treatment has remained mostly unchanged for more than 30 years, with a poor 16.4% 5-year survival. This poor survival is caused by the limitation of the 'one-dose-fits-all' paradigm which neglects the diverse spectrum of clinical presentation in NSCLC. To improve the treatment, my group and I will harness the capacities of novel cutting-edge artificial intelligence techniques combined with a massive retrospective database of patients data to answer a fundamental question about lung cancer which is "How will the disease progress?". More precisely, the deep learning approach will be used to extract general trends relating patient's data features (histopathology, anatomy, tumour stage, tumour activity, treatment plan) to an outcome (death, recurrence, secondary fibrosis, heart failure and success). The methodology output will then be used for two endpoints of the study. It will first be directly used to inform and personalise the radiotherapy treatment planning strategy to improve patient survival. It will also serve as a basis to define a new stratification procedure for lung cancer patients to refine the clinical trials selection system. This framework will enact a paradigm change in treatment planning for radiotherapy and has the potential to enable a jump in performance of the treatment by tailoring the dose to the patient; thereby lowering the secondary effects and improving overall survival.

The Neuromod+ network will represent UK research, industry, clinical and patient communities, working together to address the challenge of minimally invasive treatments for brain disorders. Increasingly, people suffer from debilitating and intractable neurological conditions, including neurodegenerative diseases and mental health disorders. Neurotechnology is playing an increasingly important part in solving these problems, leading to recent bioelectronic treatments for depression and dementia. However, the invasiveness of existing approaches limits their overall impact. Neuromod+ will bring together neurotechnology stakeholders to focus on the co-creation of next generation, minimally invasive brain stimulation technologies. The network will focus on transformative research, new collaborations, and facilitating responsible innovation, partnering with bioethicists and policy makers. As broadening the accessibility of brain modification technology my lead to unintended consequences, considering the ethical and societal implications of these technological development is of the utmost importance, and thus we will build in bioethics research as a core network activity. The activities of NEUROMOD+ will have global impact, consolidating the growing role of UK neurotechnology sector.

Delivery of pathology tissue diagnoses, most of which are cancer, in the current format is unsustainable. Advances in genomic medicine and immune-oncology have shown that the classification of tumours into subtypes allows selection of patients for specific treatments but also spares patients unnecessary toxic and expensive therapies. Still, making such diagnoses has become more time-consuming, involving the selection and interpretation of ancillary tests which requires an ever-growing specialist knowledge for each cancer type. Whilst the need for diagnostic expertise is increasing, there is already a shortfall of 25% of pathologists who are able to report results: this is set to decline. We propose that the use of AI can ensure that the delivery of tissue diagnoses by pathologists is sustainable and supports delivery of personalised treatments. The benefits of AI in pathology are beginning to be seen, e.g. identification of high-grade areas of prostate cancer shows a reduction in errors and pathologists' time. The development of AI for diagnoses is timely as full adoption of digitised histological images, allowing them to be interrogated by both humans and artificial intelligence (AI), is expected in the UK by 2025. AI is a data-hungry process; it is unrealistic to provide 100,000s images that are required to train a model. Even the most common cancers (e.g. breast) have multiple subtypes; identification of these is required for selection of patients for personalised treatments. To address this challenge, we propose to develop a novel AI strategy using a relatively small sample size (~1000 images per class). Such a model could be adapted to any cancer type. A multiple-instance learning framework will be developed, using transformers for feature extraction and classification. A tool that flags samples that cannot be confidently classified will be applied thereby alerting the pathologist of potentially unseen diseases. The deep learning model will be strengthened by the injection of pathologists' domain knowledge. Soft tissue and bone tumours We will develop the AI model on tumours of soft tissue (muscle, fat, blood vessels, etc.) and bone, an area considered to be one of the most challenging diagnostically. These tumours comprise approximately 100 different subtypes, and represent some of the most common cancers in children and young adults. We will build on our existing deep learning model of 15 different subtypes trained on 2122 images, which predicts the correct diagnosis in 87% of cases. Selection of confirmatory ancillary tests is then prompted by the algorithm and streamlines the diagnostic pathway. 17,000 images that have already been scanned will be added to the library and allow the rapid development and extension of the classification model. The image library will be linked to clinical outcomes and expanded to 35,000 images during the project. Added to this is the commitment of the established Sarcoma Network of at least 20 pathologists from across all countries in the UK, to provide the additional 20,000 images mentioned above. Additional benefits The study and infrastructure will serve as the framework for the continued development of the model which can rapidly be expanded prospectively with the introduction of digital pathology in the NHS and globally. The model can be developed over time in response to new advances. The image library will be available for training future pathologists, research, validation of other AI algorithms, and contribute to the Sarcoma Genomics England Clinical Interpretation Partnership (GeCIP) offering a valuable resource for future multi-modal multi-omic research. Working closely with Sarcoma charities, and partners, we will involve and engage patients, their families, and the public, to build trust in the use of AI in health care. Development of AI models for digitised pathology images can avert the crisis facing this medical specialty.

Aims and Background to the research Hearing loss is the most common sensory disorder in humans with 1.5 billion people affected by this during their lifetime. Some types of hearing loss are more common as we get older and therefore the burden of hearing loss is predicted to rise further as our aging population increases. Hearing loss has a significant impact on quality of life, patient health and safety as well as placing a huge demand on increasingly stretched public health services. In order to prepare and respond to this rapidly accelerating public health crisis we need to better understand the different types of hearing loss to identify who is most at risk of both worsening hearing loss but also other associated medical conditions and also which patients are most likely to benefit from new treatments. The digitialisation of patient health records offers an exciting opportunity to use the latest advances in computer science methods to look at large amounts of hearing health patient data and answer these questions. Another key part of this project will look at clinical photographs of the ear drum. Access to trained specialists who can assess the appearance of ear drums is limited in the community and there are often long waits for referrals to specialty ENT services. This situation is even worse globally in resource-poor countries. To address this problem, we propose to develop an automated programme that analyse photographs of the ear drum. The clinical images will also be used to assess whether changes in the ear drum could signal the presence of vascular disease and diabetes, much like retinal screening is performed in the eye. The ear is a more readily accessible area than the eye and could provide an easy and cost-effective site for screening. Methods We will create a store of patient data that has been collected routinely as part of standard NHS clinical care. This will include demographic details, test results, measurements, and details of medical conditions. A powerful computer programme will be used to analyse this data and describe different types of hearing loss as well as how these hearing loss types change over time. We will perform further analysis to identify links between these hearing loss subtypes and other medical conditions including dementia, diabetes, stroke and high blood pressure. For the second part of this study, we will use pictures of ear drums captured by a new medical device to develop and train a computer programme that can identify and grade the key components of an ear drum that are assessed by ENT specialists. We will use this programme alongside supplied patient details to explore whether there are changes in the ear drum that can predict the presence of diabetes and heart disease. Anticipated Outcomes The key aim of this research is to better understand the natural history of hearing loss. Identifying patients who are at higher risk of developing severe hearing loss is important for resource planning, patient counselling and identifying people who are most likely to benefit from emerging treatments or clinical trials. Identifying new associations between hearing loss and other conditions could identify patients who are "at risk" prompting earlier diagnosis and act as an opportunity for early intervention and the promotion of lifestyle modifications to divert or delay the onset of such conditions through behavioural change.