Collaborative Agentic AI for Personalized Treatment Protocol Optimization: Autonomous Clinical Decision Networks

What's New in Spine Surgery.

The autonomous business decision systems are now the centre of organisational competitiveness as organisations continue to work in volatile and increasingly data-intensive and interdependent environments. The use of multi-agent systems (MAS) can provide an effective paradigm, which facilitates decentralised, adaptive and collaborative intelligence with distributed agents perceiving environments and negotiating actions and autonomous optimisation. This paper is a critical assessment of how multi-agent AI models, such as logic agents, multi-agent reinforcement learning (MARL), agents based on large language models, agent development systems, simulation systems, and self-evolving agents' systems, have contributed to autonomous business decision systems. Systematic literature reviews, simulations, enterprise case studies and foundation-model based research and decision intelligence (DI) frameworks evidence shows how multi agent strategies provides increased robustness, scalability and responsiveness to the complex organisational domains. Simulation-based research using MAS is seen to be able to model emerging behaviours, test organisational conditions and uncertainty within energy system, construction, logistics and financial fields. The further reinforcement learning and cooperative decision modelling enhance MAS autonomy as it provides the ability to optimise the learning of dynamic environments. Similar results are presented by the research of Automated Machine Learning (AutoML) that demonstrates that automated model development helps lower technical barriers and speeds up its deployment and enables adaptive decision pipelines. Decision Intelligence literature displays how organisations can restructure decision-making with the help of AI to construct data pipelines, and provide human-computer interaction. Although issues remain, such as explainability, cost of coordination and data governance, there is strong evidence that MAS constitute a backbone architecture to next generation autonomous decision ecosystems according to all sources. The given synthesis illustrates how multi-agent AI may be used as a strategic foundation of predictive, operational and strategic automation of decisions in contemporary businesses.

Hybrid beamforming with massive multiple-input multiple-output (MIMO) antenna arrays plays a pivotal role in achieving high data rates and spectral efficiency in next-generation wireless systems. Nevertheless, the real-time 62design and adaptation of beamforming codebooks and user grouping remain challenging, as conventional machine learning (ML) methods typically rely on the centralized collection of extensive channel state information (CSI), resulting in substantial feedback overhead and potential privacy violations. This chapter proposes a federated multi-agent reinforcement learning (MARL) framework for joint hybrid beamforming and dynamic user grouping. Unlike prior work which focuses solely on centralized codebook learning, the proposed approach enables distributed user equipment (UE) to collaboratively train codebook and clustering algorithms without sharing raw data, thereby preserving privacy. The proposed method significantly reduces communication overhead while achieving near-optimal beam patterns and robust scalability in dense network scenarios, making it well-suited for practical deployment in next-generation wireless systems. The experimental findings indicate that for mobile users, federated learning (FL)-based user clustering with MARL outperforms traditional initial access-based beam clustering in MARL.

As cyber threats become increasingly automated and sophisticated, novel solutions must be introduced to improve defense of enterprise networks. Deep reinforcement learning (DRL) has demonstrated potential in mitigating these advanced threats. Single DRL agents have proven utility toward execution of autonomous cyber defense. Despite the success of employing single DRL agents, this approach presents significant limitations, especially regarding scalability within large enterprise networks. An attractive alternative to the single‐agent approach is the use of multi‐agent reinforcement learning (MARL). However, developing MARL agents is costly with few options for examining MARL cyber defense techniques against adversarial agents. This paper presents a MARL network security environment, the fourth iteration of the cyber autonomy gym for experimentation (CAGE) challenges. This challenge was specifically designed to test the efficacy of MARL algorithms in an enterprise network. Our work aims to evaluate the potential of MARL as a robust and scalable solution for autonomous network defense.

As cyber threats become increasingly automated and sophisticated, novel solutions must be introduced to improve defence of enterprise networks. Deep Reinforcement Learning (DRL) has demonstrated potential in mitigating these advanced threats. Single DRL Agents have proven utility toward execution of autonomous cyber defence. Despite the success of employing single DRL Agents, this approach presents significant limitations, especially regarding scalability within large enterprise networks. An attractive alternative to the single agent approach is the use of Multi-Agent Reinforcement Learning (MARL). However, developing MARL agents is costly with few options for examining MARL cyber defence techniques against adversarial agents. This paper presents a MARL network security environment, the fourth iteration of the Cyber Autonomy Gym for Experimentation (CAGE) challenges. This challenge was specifically designed to test the efficacy of MARL algorithms in an enterprise network. Our work aims to evaluate the potential of MARL as a robust and scalable solution for autonomous network defence.

Long-term, flexible therapy strategies are needed for chronic diseases like cardiovascular ailments, diabetes, and chronic kidney disease and they have to deal with patient heterogeneity, changing physiological states and privacy constraints. This research suggests a framework for privacy-preserving and personalized artificial intelligence that combines Federated Learning (FL), a Res-HyperTransformerNet deep prediction model, Personalized Multi-Agent Reinforcement Learning (PMARL) and Explainable Artificial Intelligence (XAI) for adaptive chronic disease management. The framework is structured with two main aims: (i) to create an accurate and privacy-aware predictive model for chronic disease risk using diverse data sources, and (ii) to permit adaptive, individual therapy optimization through multi-agent reinforcement learning. Federated learning is utilized to conduct the training of Res-HyperTransformerNet over distributed Internet of Medical Things (IoMT) nodes without the requirement of transferring raw patient data. The embeddings that are created from patient data are then passed on to a PMARL module, where several agents optimize therapy dimensions like medications, diet, physical activity, and mental health interventions independently. To make the clinical support more understandable, the SHAP-based explainability method is used for both predictive and decision-making parts. The framework is tested using two public datasets—the CDC Chronic Disease dataset and the UCI Chronic Kidney Disease Risk Factor dataset. The performance is measured by employing classification metrics (accuracy, precision, recall, F1-score, MCC) as well as reinforcement learning metrics (reward score, convergence steps, episode return) and federated system metrics (communication overhead, convergence rounds, and training time). The experiments indicate that the proposed framework demonstrated improved performance in terms of predictive accuracy and policy convergence speed compared to the baseline deep learning and reinforcement learning models, and at the same time it is more cost-effective when it comes to communication in a federated setting. This means that the proposed method suggests the potential of being a reproducible and open AI framework for adaptive chronic disease therapy management. The combination of ResNet and Transformer blocks, i.e., Federated Res-HyperTransformerNet, achieved strong performance in both datasets, namely CDC Chronic Disease (Dataset 1) and UCI CKD Risk Factor (Dataset 2). The obtained accuracy for Dataset 1 was 98.61% and for Dataset 2, it was 97.75%, under the conducted experimental settings.

Atraumatic shoulder instability (ASI) poses a significant challenge in orthopedic practice, characterized by functional and anatomical deficits in the glenohumeral joint (GHJ) resulting from chronic overuse rather than acute trauma. This review aims to provide a comprehensive summary of the current understanding of ASI, covering its anatomy, etiology, the influence of psychosocial factors and primarily focusing on management strategies. The management of ASI is a central focus for restoring normal biomechanical and anatomical function of the injured upper extremity. A literature review was conducted, seeking to further understand the current treatment approaches of this condition. The management of ASI employs differing combinations of physiotherapy, rehabilitation and surgery, with treatment dictated by patient stratification models and patient reported outcome measures (PROMs). Clinical stratification of patients along a diagnostic continuum aid devising appropriate patient-centered treatment and is based on the extent of anatomical disruption. Exercise-focused therapy, particularly the Watson Instability Program, is highlighted as a promising approach, emphasizing scapular and rotator cuff muscle strengthening to restore active shoulder control. Surgical interventions, such as open inferior capsular shift, arthroscopic capsular plication, and electrothermal arthroscopic capsulorrhaphy, are discussed in detail, with emphasis on their indications, outcomes, and comparative effectiveness. We also shed a light on the psychosocial factors role in ASI management outcomes, underscoring the importance of a multidisciplinary approach in patient care. In summary, ASI management necessitates a holistic approach that integrates anatomical understanding, tailored treatment protocols, and consideration of psychosocial factors. Further research is warranted to refine treatment strategies, validate rehabilitation programs, and explore the interplay between psychological factors and physical outcomes in ASI patients.

Background As the treatment paradigm of inflammatory bowel disease changes towards patient-centred treatment, it is becoming increasingly important to measure patient-reported outcomes (PROs). We aimed to identify the changes of emotional health and work- or activity-related outcomes one year after the diagnosis of ulcerative colitis (UC) and its predictors in patients enrolled in the moderate-to-severe UC in Korea (MOSAIK) cohort (ClinicalTrials.gov: NCT02229344). Methods The MOSAIK cohort is the first nationwide, prospective, inception cohort on moderate-to-severe UC in Korea. Between August 2014 and March 2017, consecutive patients from 30 tertiary hospitals were enrolled. PRO data including hospital anxiety and depression scale (HADS) for emotional health, and work productivity and activity impairment (WPAI) questionnaire for work- or activity-related outcomes, were collected within the first 4 weeks of diagnosis via patient surveys. Wilcoxon-signed rank tests and linear mixed-effects regression models were used for paired comparisons between baseline and 1 year and assessing the predictors of HADS and WPAI. Results Of the 368 enrolled patients, 333 eligible patients were analyzed. The mean age at diagnosis was 39 years and 57.7% (192/333) were male. A considerable number of patients had moderate to high (≥11 by HADS) levels of anxiety and depression (16.0% and 20.5%, respectively), and about half of patients had work and activity impairment (45.5% and 45.8%%, respectively) at baseline. After 1 year follow-up, significant reduction of anxiety and depression (mean difference [MD] in HADS score -1.3 for both anxiety and depression, P<0.001), as well as work and activity impairment (MD -24.1% and -22.4%, P<0.001) was noted. Higher disease activity (partial Mayo Score) during a one-year period was a significant predictor of anxiety, depression, work and social activity impairment. Among the symptoms of UC, abdominal pain was a significant predictor of depression and work and activity impairment, and weight loss and diarrhoea were significant predictors of activity impairment. Conclusion Newly diagnosed moderate-to-severe UC patients had considerable anxiety, depression, and work and activity impairment at baseline, but significant improvement was noted after 1 year. Controlling symptoms and disease activity was the most important factor to improve PROs after 1 year.

Opportunities for Improving Glaucoma Clinical Trials via Deep Learning-Based Identification of Patients with Low Visual Field Variability

Patient-reported outcomes (PROs) are often used by clinicians to evaluate patient response to specialty medications used to treat multiple sclerosis (MS) and rheumatologic conditions. Identifying associations among PROs and patient characteristics could inform patient-centered treatment monitoring. To examine the association among patient characteristics and PROs, including patient-reported adherence (defined as no missed doses), medication tolerance, patient perceived effectiveness, and health care resource utilization (HCRU; defined as emergency department visits or hospitalizations), for patients prescribed specialty medications in 2 health system specialty pharmacies. A dual-center, retrospective review of monthly medication assessments completed by Vanderbilt Specialty Pharmacy and University of Illinois Hospital and Health Sciences System specialty pharmacy was conducted. Patients were included if they received at least 3 fills of a specialty medication from rheumatology or MS clinics from October 2019 to March 2022, excluding patients with more than a 30-day supply. Primary outcomes were the PROs of patient-reported adherence, medication tolerability, perceived effectiveness, and HCRU. For each of the 2 primary outcomes (adherence and tolerability), a mixed-effects logistic regression model was used to test for associations with age, sex, race, clinic, site, and the other PROs. A total of 61,926 assessments were completed from 3,677 patients (Site 1 = 3,346; 91.0% and Site 2 = 331; 9.0%). Patients were predominantly White (75.6%) and female (71.7%) with a median age of 50 years (IQR = 37-61). Assessments most frequently originated from rheumatology (76.0%). Nonadherence was reported 4.0% of the time, with the most common explanations being forgetfulness (33.1%) and medication being held because of a procedure or illness (29.5%). Most responses indicated perceived effectiveness as good/excellent (93.9%), with 98.5% of responses indicating no issues with tolerability. Patients who reported tolerability issues were 2.5 times more likely to report a missed dose (95% CI = 1.87-3.23, P < 0.001). An effectiveness rating of fair was associated with a 61% increase in the odds of a missed dose compared with a rating of good/excellent (95% CI = 1.33-1.94). Patients filling rheumatology or MS specialty medications within health system specialty pharmacies reported high rates of medication effectiveness and adherence and low rates of issues with tolerability and HCRU. Patients who report tolerability issues or lower perceived effectiveness may benefit from additional monitoring to prevent nonadherence.

Autonomous vehicles (AVs) are expected to operate safely and efficiently in complex urban environments characterized by dynamic and uncertain elements such as pedestrians, cyclists and adverse weather. Although current neural network-based decision-making algorithms, fuzzy logic and reinforcement learning have shown promise, they often struggle to handle ambiguous situations, such as partially hidden road signs or unpredictable human behavior. This paper proposes a new hybrid decision-making framework combining multi-agent reinforcement learning (MARL) and explainable artificial intelligence (XAI) to improve robustness, adaptability and transparency. Each agent of the MARL architecture is specialized in a specific sub-task (e.g., obstacle avoidance, trajectory planning, intention prediction), enabling modular and cooperative learning. XAI techniques are integrated to provide interpretable rationales for decisions, facilitating human understanding and regulatory compliance. The proposed system will be validated using CARLA simulator, combined with reference data, to demonstrate improved performance in safety-critical and ambiguous driving scenarios.

Inequity in diabetes-related health due to socioeconomic status is a recognised challenge. A patient-centred approach tailors care to individual needs, and patient-reported outcome (PRO) measures can be a valuable tool in this process. In Denmark, a new diabetes-specific PRO questionnaire has been developed to enable a systematic assessment of patients' experiences and potential barriers to care. This study examined participation in the PRO diabetes questionnaire and its relation to socioeconomic status and severe psychiatric comorbidity. This register study included people with diabetes who were invited to complete the questionnaire. Participation status was analysed using multiple logistic regression models, incorporating variables related to socioeconomic status. Socioeconomic status was a significant overall predictor of questionnaire participation, with non-Western immigration showing the strongest association. Additionally, Western immigration, low educational attainment and severe psychiatric comorbidity were significant predictors in subsets of the regression models. This study highlights disparities in participation in the PRO diabetes questionnaire related to socioeconomic status and severe psychiatric comorbidity. To promote equitable access and reduce diabetes-related health inequalities, targeted efforts are needed to support vulnerable groups in engaging with patient-centred interventions. Steno Diabetes Center Copenhagen TRIAL REGISTRATION. Pactius.

Patient-reported outcome measures (PROMs) are increasingly applied in clinical practice to improve quality of care and patient outcomes. To date, however, little is known about their use for peer comparisons of performance between clinicians, which the authors refer to as benchmarking. In this study of cataract surgeons, benchmarking was implemented through a stepwise approach: (1) Before benchmarking, each surgeon could not access their performance data or compare it with that of their peers; (2) once benchmarking began, all surgeons’ identities and individual results became visible to their peers. This benchmarking design was intended to create a learning community free from judgment or stigmatization, drawing each surgeon into a collaborative dynamic of continuous improvement. The authors hypothesized that surgeons within such a benchmarking community would be encouraged to modify their surgical indications, improve their patients’ average outcomes, and reduce the volume of procedures without meaningful patient benefit. Between 2021 and 2025, the authors analyzed a benchmarking community of cataract surgeons practicing at four institutions in France. The benchmarking model included voluntary participation, patient involvement, improvement cycles, and respect for patient–surgeon autonomy in clinical decision-making. Dashboards allowed surgeons to compare their PROMs, clinician-reported outcome measures (CROMs), and case-mix–adjusted results. Access to the benchmarking dashboard was supplemented through noncoercive, nonpunitive quarterly meetings with peers to discuss individual and community results, enabling improvement cycles. The authors analyzed data from over 2600 patients with complete PROM and CROM data and found that five times more surgeons modified the surgical indications they used during benchmarking than before benchmarking — 10 of 24 (42%) versus 2 of 24 (8%). The authors also found that only surgeons who actively consulted the benchmarking dashboard modified their surgical indications. Among those who modified their surgical indications during benchmarking, the average postsurgery health gain score of their patients increased by 22% (from 28.7 before to 35.1 after modifying the indications), and the rate of patients without meaningful benefit decreased by 34% (from 19% to 13%). The differences were similar when compared with performance before benchmarking, which yielded a 21% increase in average health gain scores and a 35% decrease in the proportion of patients below the meaningful outcome threshold. These findings support the potential of unmasked peer benchmarking to improve patient outcomes, reduce unnecessary procedures, and advance value-based surgical care.

This paper presents an integrated approach to real-time processing and analysis of large-scale data streams using artificial intelligence methods, multi-agent systems, and multi-layered control architectures. Particular attention is devoted to combining technologies that are traditionally applied separately: unmanned aerial vehicle (UAV) control systems, situational centers, streaming video services, and distributed sensor networks. The proposed model accounts for the stochastic nature of the environment, the dynamic behaviour of data flows, and resource constraints, which are critical factors for real-time systems. The study develops a multi-level multi-agent architecture that includes central, regional, and edge nodes, as well as local agents such as drones, sensors, and client-side video devices. A mathematical formalization of agent interaction and reward functions is proposed to ensure a balance between service quality, latency, packet loss, and energy consumption. The use of multi-agent reinforcement learning (MARL) algorithms is introduced to support adaptive decision-making in real time, along with the development of a digital twin of the environment for predicting future system states using deep and generative models. The results demonstrate that the integration of edge computing, hierarchical decision-making, and local agent autonomy significantly enhances system resilience, provides stable performance under external disturbances, and enables effective infrastructure scaling. The findings form a scientific foundation for designing adaptive, robust, and self-learning next-generation digital ecosystems capable of intelligent processing of large data streams in real time

A pragmatic randomized clinical trial of multilevel interventions to improve adherence to lung cancer screening (The Larch Study): Study protocol