PET/CT based transformer model for multi-outcome prediction in oropharyngeal cancer.

BACKGROUND AND PURPOSE: To optimize our previously proposed TransRP, a model integrating CNN (convolutional neural network) and ViT (Vision Transformer) designed for recurrence-free survival prediction in oropharyngeal cancer and to extend its application to the prediction of multiple clinical outcomes, including locoregional control (LRC), Distant metastasis-free survival (DMFS) and overall survival (OS). MATERIALS AND METHODS: Data was collected from 400 patients (300 for training and 100 for testing) diagnosed with oropharyngeal squamous cell carcinoma (OPSCC) who underwent (chemo)radiotherapy at University Medical Center Groningen. Each patient's data comprised pre-treatment PET/CT scans, clinical parameters, and clinical outcome endpoints, namely LRC, DMFS and OS. The prediction performance of TransRP was compared with CNNs when inputting image data only. Additionally, three distinct methods (m1-3) of incorporating clinical predictors into TransRP training and one method (m4) that uses TransRP prediction as one parameter in a clinical Cox model were compared. RESULTS: TransRP achieved higher test C-index values of 0.61, 0.84 and 0.70 than CNNs for LRC, DMFS and OS, respectively. Furthermore, when incorporating TransRP's prediction into a clinical Cox model (m4), a higher C-index of 0.77 for OS was obtained. Compared with a clinical routine risk stratification model of OS, our model, using clinical variables, radiomics and TransRP prediction as predictors, achieved larger separations of survival curves between low, intermediate and high risk groups. CONCLUSION: TransRP outperformed CNN models for all endpoints. Combining clinical data and TransRP prediction in a Cox model achieved better OS prediction.

Effect of emphysema on AI software and human reader performance in lung nodule detection from low-dose chest CT.

BACKGROUND: Emphysema influences the appearance of lung tissue in computed tomography (CT). We evaluated whether this affects lung nodule detection by artificial intelligence (AI) and human readers (HR). METHODS: Individuals were selected from the "Lifelines" cohort who had undergone low-dose chest CT. Nodules in individuals without emphysema were matched to similar-sized nodules in individuals with at least moderate emphysema. AI results for nodular findings of 30-100 mm3 and 101-300 mm3 were compared to those of HR; two expert radiologists blindly reviewed discrepancies. Sensitivity and false positives (FPs)/scan were compared for emphysema and non-emphysema groups. RESULTS: Thirty-nine participants with and 82 without emphysema were included (n = 121, aged 61 ± 8 years (mean ± standard deviation), 58/121 males (47.9%)). AI and HR detected 196 and 206 nodular findings, respectively, yielding 109 concordant nodules and 184 discrepancies, including 118 true nodules. For AI, sensitivity was 0.68 (95% confidence interval 0.57-0.77) in emphysema versus 0.71 (0.62-0.78) in non-emphysema, with FPs/scan 0.51 and 0.22, respectively (p = 0.028). For HR, sensitivity was 0.76 (0.65-0.84) and 0.80 (0.72-0.86), with FPs/scan of 0.15 and 0.27 (p = 0.230). Overall sensitivity was slightly higher for HR than for AI, but this difference disappeared after the exclusion of benign lymph nodes. FPs/scan were higher for AI in emphysema than in non-emphysema (p = 0.028), while FPs/scan for HR were higher than AI for 30-100 mm3 nodules in non-emphysema (p = 0.009). CONCLUSIONS: AI resulted in more FPs/scan in emphysema compared to non-emphysema, a difference not observed for HR. RELEVANCE STATEMENT: In the creation of a benchmark dataset to validate AI software for lung nodule detection, the inclusion of emphysema cases is important due to the additional number of FPs. KEY POINTS: • The sensitivity of nodule detection by AI was similar in emphysema and non-emphysema. • AI had more FPs/scan in emphysema compared to non-emphysema. • Sensitivity and FPs/scan by the human reader were comparable for emphysema and non-emphysema. • Emphysema and non-emphysema representation in benchmark dataset is important for validating AI.

An in silico modeling approach to understanding the dynamics of the post-burn immune response.

Introduction: Burns are characterized by a massive and prolonged acute inflammation, which persists for up to months after the initial trauma. Due to the complexity of the inflammatory process, Predicting the dynamics of wound healing process can be challenging for burn injuries. The aim of this study was to develop simulation models for the post-burn immune response based on (pre)clinical data. Methods: The simulation domain was separated into blood and tissue compartments. Each of these compartments contained solutes and cell agents. Solutes comprise pro-inflammatory cytokines, anti-inflammatory cytokines and inflammation triggering factors. The solutes diffuse around the domain based on their concentration profiles. The cells include mast cells, neutrophils, and macrophages, and were modeled as independent agents. The cells are motile and exhibit chemotaxis based on concentrations gradients of the solutes. In addition, the cells secrete various solutes that in turn alter the dynamics and responses of the burn wound system. Results: We developed an Glazier-Graner-Hogeweg method-based model (GGH) to capture the complexities associated with the dynamics of inflammation after burn injuries, including changes in cell counts and cytokine levels. Through simulations from day 0 - 4 post-burn, we successfully identified key factors influencing the acute inflammatory response, i.e., the initial number of endothelial cells, the chemotaxis threshold, and the level of chemoattractants. Conclusion: Our findings highlight the pivotal role of the initial endothelial cell count as a key parameter for intensity of inflammation and progression of acute inflammation, 0 - 4 days post-burn.

Mild obstructive sleep apnoea in females: post hoc analysis of the MERGE randomised controlled trial.

Background: A post hoc analysis of the MERGE trial was conducted, to investigate whether sex differences are evident at the mildest end of the disease spectrum, for symptoms associated with obstructive sleep apnoea (OSA) and the response to continuous positive airway pressure (CPAP) treatment. Methods: MERGE participants with mild OSA (apnoea-hypopnoea index 5-15 events·h-1; American Academy of Sleep Medicine 2012 criteria) were randomised to either CPAP plus standard care (sleep hygiene counselling) or standard care alone for 3 months. Quality of life (QoL) was measured by questionnaires completed before and after the 3 months. This post hoc analysis of participants of the MERGE trial compared the symptom presentation, and response to CPAP, between the sexes. Results: 233 patients were included; 71 (30%) were female. Females were more symptomatic at baseline in all QoL questionnaires. Specifically, females had lower 36-item Short-Form Health Survey (SF-36) Vitality scores (mean±sd 39.1±10.1 versus 44.8±10.3) and higher Epworth Sleepiness Scale (ESS) scores (mean±sd 11.0±4.2 versus 9.5±4.4). Both sexes experienced snoring, but more females reported fatigue and more males reported witnessed apnoeas. All symptoms improved with CPAP for both sexes; however, females had larger improvements in SF-36 Vitality scores, which was the primary outcome of the MERGE trial (mean change 9.4 (95% CI 6.8-12.0) versus 6.0 (95% CI 4.3-7.7); p=0.034), and ESS (mean change -4.1 (95% CI -5.1- -3.0) versus -2.5 (95% CI -3.1- -1.8); p=0.015), after adjustment for baseline scores and CPAP usage. Conclusions: Sex differences are apparent in patients with mild OSA. Females experience worse QoL symptoms than males at presentation to the sleep clinic; however, these improve significantly with CPAP treatment.

Protocol for analyzing emergence dynamics of diabetes with obesity using numerical continuation and bifurcation analysis.

Type 2 diabetes (T2D) is a multifactorial disease that slowly and inconspicuously progresses over years. Here, we present a protocol for analyzing slow progression dynamics of T2D with obesity. We describe steps for using software to exploit the differences between the timescales of the metabolic variables and using numerical continuation and bifurcation analysis. We detail procedures to analyze bi-stable system dynamics and identify the thresholds that separate healthy and diabetic states. For complete details on the use and execution of this protocol, please refer to Yildirim et al. (2023).1.

Structural Predictors of Lung Function Decline in Young Smokers with Normal Spirometry.

Rationale: Chronic obstructive pulmonary disease (COPD) due to tobacco smoking commonly presents when extensive lung damage has occurred. Objectives: We hypothesized that structural change would be detected early in the natural history of COPD and would relate to loss of lung function with time. Methods: We recruited 431 current smokers (median age, 39 yr; 16 pack-years smoked) and recorded symptoms using the COPD Assessment Test (CAT), spirometry, and quantitative thoracic computed tomography (QCT) scans at study entry. These scan results were compared with those from 67 never-smoking control subjects. Three hundred sixty-eight participants were followed every six months with measurement of postbronchodilator spirometry for a median of 32 months. The rate of FEV1 decline, adjusted for current smoking status, age, and sex, was related to the initial QCT appearances and symptoms, measured using the CAT. Measurements and Main Results: There were no material differences in demography or subjective CT appearances between the young smokers and control subjects, but 55.7% of the former had CAT scores greater than 10, and 24.2% reported chronic bronchitis. QCT assessments of disease probability-defined functional small airway disease, ground-glass opacification, bronchovascular prominence, and ratio of small blood vessel volume to total pulmonary vessel volume were increased compared with control subjects and were all associated with a faster FEV1 decline, as was a higher CAT score. Conclusions: Radiological abnormalities on CT are already established in young smokers with normal lung function and are associated with FEV1 loss independently of the impact of symptoms. Structural abnormalities are present early in the natural history of COPD and are markers of disease progression. Clinical trial registered with www.clinicaltrials.gov (NCT03480347).

Localization of contrast-enhanced breast lesions in ultrafast screening MRI using deep convolutional neural networks.

OBJECTIVES: To develop a deep learning-based method for contrast-enhanced breast lesion detection in ultrafast screening MRI. MATERIALS AND METHODS: A total of 837 breast MRI exams of 488 consecutive patients were included. Lesion's location was independently annotated in the maximum intensity projection (MIP) image of the last time-resolved angiography with stochastic trajectories (TWIST) sequence for each individual breast, resulting in 265 lesions (190 benign, 75 malignant) in 163 breasts (133 women). YOLOv5 models were fine-tuned using training sets containing the same number of MIP images with and without lesions. A long short-term memory (LSTM) network was employed to help reduce false positive predictions. The integrated system was then evaluated on test sets containing enriched uninvolved breasts during cross-validation to mimic the performance in a screening scenario. RESULTS: In five-fold cross-validation, the YOLOv5x model showed a sensitivity of 0.95, 0.97, 0.98, and 0.99, with 0.125, 0.25, 0.5, and 1 false positive per breast, respectively. The LSTM network reduced 15.5% of the false positive prediction from the YOLO model, and the positive predictive value was increased from 0.22 to 0.25. CONCLUSIONS: A fine-tuned YOLOv5x model can detect breast lesions on ultrafast MRI with high sensitivity in a screening population, and the output of the model could be further refined by an LSTM network to reduce the amount of false positive predictions. CLINICAL RELEVANCE STATEMENT: The proposed integrated system would make the ultrafast MRI screening process more effective by assisting radiologists in prioritizing suspicious examinations and supporting the diagnostic workup. KEY POINTS: • Deep convolutional neural networks could be utilized to automatically pinpoint breast lesions in screening MRI with high sensitivity. • False positive predictions significantly increased when the detection models were tested on highly unbalanced test sets with more normal scans. • Dynamic enhancement patterns of breast lesions during contrast inflow learned by the long short-term memory networks helped to reduce false positive predictions.

Masked Conditional Variational Autoencoders for Chromosome Straightening.

Karyotyping is of importance for detecting chromosomal aberrations in human disease. However, chromosomes easily appear curved in microscopic images, which prevents cytogeneticists from analyzing chromosome types. To address this issue, we propose a framework for chromosome straightening, which comprises a preliminary processing algorithm and a generative model called masked conditional variational autoencoders (MC-VAE). The processing method utilizes patch rearrangement to address the difficulty in erasing low degrees of curvature, providing reasonable preliminary results for the MC-VAE. The MC-VAE further straightens the results by leveraging chromosome patches conditioned on their curvatures to learn the mapping between banding patterns and conditions. During model training, we apply a masking strategy with a high masking ratio to train the MC-VAE with eliminated redundancy. This yields a non-trivial reconstruction task, allowing the model to effectively preserve chromosome banding patterns and structure details in the reconstructed results. Extensive experiments on three public datasets with two stain styles show that our framework surpasses the performance of state-of-the-art methods in retaining banding patterns and structure details. Compared to using real-world bent chromosomes, the use of high-quality straightened chromosomes generated by our proposed method can improve the performance of various deep learning models for chromosome classification by a large margin. Such a straightening approach has the potential to be combined with other karyotyping systems to assist cytogeneticists in chromosome analysis.

Deep learning-based outcome prediction using PET/CT and automatically predicted probability maps of primary tumor in patients with oropharyngeal cancer.

BACKGROUND AND OBJECTIVE: Recently, deep learning (DL) algorithms showed to be promising in predicting outcomes such as distant metastasis-free survival (DMFS) and overall survival (OS) using pre-treatment imaging in head and neck cancer. Gross Tumor Volume of the primary tumor (GTVp) segmentation is used as an additional channel in the input to DL algorithms to improve model performance. However, the binary segmentation mask of the GTVp directs the focus of the network to the defined tumor region only and uniformly. DL models trained for tumor segmentation have also been used to generate predicted tumor probability maps (TPM) where each pixel value corresponds to the degree of certainty of that pixel to be classified as tumor. The aim of this study was to explore the effect of using TPM as an extra input channel of CT- and PET-based DL prediction models for oropharyngeal cancer (OPC) patients in terms of local control (LC), regional control (RC), DMFS and OS. METHODS: We included 399 OPC patients from our institute that were treated with definitive (chemo)radiation. For each patient, CT and PET scans and GTVp contours, used for radiotherapy treatment planning, were collected. We first trained a previously developed 2.5D DL framework for tumor probability prediction by 5-fold cross validation using 131 patients. Then, a 3D ResNet18 was trained for outcome prediction using the 3D TPM as one of the possible inputs. The endpoints were LC, RC, DMFS, and OS. We performed 3-fold cross validation on 168 patients for each endpoint using different combinations of image modalities as input. The final prediction in the test set (100) was obtained by averaging the predictions of the 3-fold models. The C-index was used to evaluate the discriminative performance of the models. RESULTS: The models trained replacing the GTVp contours with the TPM achieved the highest C-indexes for LC (0.74) and RC (0.60) prediction. For OS, using the TPM or the GTVp as additional image modality resulted in comparable C-indexes (0.72 and 0.74). CONCLUSIONS: Adding predicted TPMs instead of GTVp contours as an additional input channel for DL-based outcome prediction models improved model performance for LC and RC.

Using network analysis to identify leverage points based on causal loop diagrams leads to false inference.

Network analysis is gaining momentum as an accepted practice to identify which factors in causal loop diagrams (CLDs)-mental models that graphically represent causal relationships between a system's factors-are most likely to shift system-level behaviour, known as leverage points. This application of network analysis, employed to quantitatively identify leverage points without having to use computational modelling approaches that translate CLDs into sets of mathematical equations, has however not been duly reflected upon. We evaluate whether using commonly applied network analysis metrics to identify leverage points is justified, focusing on betweenness- and closeness centrality. First, we assess whether the metrics identify the same leverage points based on CLDs that represent the same system but differ in inferred causal structure-finding that they provide unreliable results. Second, we consider conflicts between assumptions underlying the metrics and CLDs. We recognise six conflicts suggesting that the metrics are not equipped to take key information captured in CLDs into account. In conclusion, using betweenness- and closeness centrality to identify leverage points based on CLDs is at best premature and at worst incorrect-possibly causing erroneous identification of leverage points. This is problematic as, in current practice, the results can inform policy recommendations. Other quantitative or qualitative approaches that better correspond with the system dynamics perspective must be explored.

Comparison of computed tomography image features extracted by radiomics, self-supervised learning and end-to-end deep learning for outcome prediction of oropharyngeal cancer.

Background and purpose: To compare the prediction performance of image features of computed tomography (CT) images extracted by radiomics, self-supervised learning and end-to-end deep learning for local control (LC), regional control (RC), locoregional control (LRC), distant metastasis-free survival (DMFS), tumor-specific survival (TSS), overall survival (OS) and disease-free survival (DFS) of oropharyngeal squamous cell carcinoma (OPSCC) patients after (chemo)radiotherapy. Methods and materials: The OPC-Radiomics dataset was used for model development and independent internal testing and the UMCG-OPC set for external testing. Image features were extracted from the Gross Tumor Volume contours of the primary tumor (GTVt) regions in CT scans when using radiomics or a self-supervised learning-based method (autoencoder). Clinical and combined (radiomics, autoencoder or end-to-end) models were built using multivariable Cox proportional-hazard analysis with clinical features only and both clinical and image features for LC, RC, LRC, DMFS, TSS, OS and DFS prediction, respectively. Results: In the internal test set, combined autoencoder models performed better than clinical models and combined radiomics models for LC, RC, LRC, DMFS, TSS and DFS prediction (largest improvements in C-index: 0.91 vs. 0.76 in RC and 0.74 vs. 0.60 in DMFS). In the external test set, combined radiomics models performed better than clinical and combined autoencoder models for all endpoints (largest improvements in LC, 0.82 vs. 0.71). Furthermore, combined models performed better in risk stratification than clinical models and showed good calibration for most endpoints. Conclusions: Image features extracted using self-supervised learning showed best internal prediction performance while radiomics features have better external generalizability.

A data-driven computational model for obesity-driven diabetes onset and remission through weight loss.

Obesity is a major risk factor for the development of type 2 diabetes (T2D), where a sustained weight loss may result in T2D remission in individuals with obesity. To design effective and feasible intervention strategies to prevent or reverse T2D, it is imperative to study the progression of T2D and remission together. Unfortunately, this is not possible through experimental and observational studies. To address this issue, we introduce a data-driven computational model and use human data to investigate the progression of T2D with obesity and remission through weight loss on the same timeline. We identify thresholds for the emergence of T2D and necessary conditions for remission. We explain why remission is only possible within a window of opportunity and the way that window depends on the progression history of T2D, individual's metabolic state, and calorie restrictions. These findings can help to optimize therapeutic intervention strategies for T2D prevention or treatment.

Automated Breast Density Assessment in MRI Using Deep Learning and Radiomics: Strategies for Reducing Inter-Observer Variability.

BACKGROUND: Accurate breast density evaluation allows for more precise risk estimation but suffers from high inter-observer variability. PURPOSE: To evaluate the feasibility of reducing inter-observer variability of breast density assessment through artificial intelligence (AI) assisted interpretation. STUDY TYPE: Retrospective. POPULATION: Six hundred and twenty-one patients without breast prosthesis or reconstructions were randomly divided into training (N = 377), validation (N = 98), and independent test (N = 146) datasets. FIELD STRENGTH/SEQUENCE: 1.5 T and 3.0 T; T1-weighted spectral attenuated inversion recovery. ASSESSMENT: Five radiologists independently assessed each scan in the independent test set to establish the inter-observer variability baseline and to reach a reference standard. Deep learning and three radiomics models were developed for three classification tasks: (i) four Breast Imaging-Reporting and Data System (BI-RADS) breast composition categories (A-D), (ii) dense (categories C, D) vs. non-dense (categories A, B), and (iii) extremely dense (category D) vs. moderately dense (categories A-C). The models were tested against the reference standard on the independent test set. AI-assisted interpretation was performed by majority voting between the models and each radiologist's assessment. STATISTICAL TESTS: Inter-observer variability was assessed using linear-weighted kappa (κ) statistics. Kappa statistics, accuracy, and area under the receiver operating characteristic curve (AUC) were used to assess models against reference standard. RESULTS: In the independent test set, five readers showed an overall substantial agreement on tasks (i) and (ii), but moderate agreement for task (iii). The best-performing model showed substantial agreement with reference standard for tasks (i) and (ii), but moderate agreement for task (iii). With the assistance of the AI models, almost perfect inter-observer variability was obtained for tasks (i) (mean κ = 0.86), (ii) (mean κ = 0.94), and (iii) (mean κ = 0.94). DATA CONCLUSION: Deep learning and radiomics models have the potential to help reduce inter-observer variability of breast density assessment. LEVEL OF EVIDENCE: 3 TECHNICAL EFFICACY: Stage 1.

A Double-Blind, Randomized, Placebo-controlled Trial of Long-Term Doxycycline Therapy on Exacerbation Rate in Patients with Stable Chronic Obstructive Pulmonary Disease.

Rationale: Chronic obstructive pulmonary disease (COPD) exacerbations are a major cause of morbidity and mortality, and preventing them is a key treatment target. Long-term macrolide treatment is effective at reducing exacerbations, but there is a paucity of evidence for other antibiotic classes. Objectives: To assess whether 12-month use of doxycycline reduces the exacerbation rate in people with COPD. Methods: People with moderate to very severe COPD and an exacerbation history were recruited from three UK centers and randomized to 12 months of doxycycline 100 mg once daily or placebo. The primary study outcome was the exacerbation rate per person-year. Results: A total of 222 people were randomized. Baseline mean FEV1 was 1.35 L (SD, 0.35 L), 52.5% predicted (SD, 15.9% predicted). The median number of treated exacerbations in the year before the study was 2 (SD, 1-4). A total of 71% of patients reported two or more exacerbations, and 81% were already prescribed inhaled corticosteroids at baseline. The COPD exacerbation rate did not differ between the groups (doxycycline/placebo rate ratio [RR], 0.86; 95% confidence interval [CI], 0.67-1.10; P = 0.23). No difference was seen if only treated exacerbations or hospitalizations were considered. In preplanned subgroup analysis, doxycycline appeared to better reduce the exacerbation rate among people with severe COPD (RR, 0.36; 95% CI, 0.15-0.85; P = 0.019) and in those with an eosinophil count <300 cells/µl (RR, 0.50; 95% CI, 0.29-0.84; P = 0.01). Health status measured by St. George's Respiratory Questionnaire was 5.2 points worse in the doxycycline group at 12 months (P < 0.007). Conclusions: Doxycycline did not significantly reduce the exacerbation rate, over 12 months, in participants with COPD who exacerbated regularly, but it may have benefitted those with more severe COPD or blood eosinophil counts <300 cells/µl. Clinical trial registered with www.clinicaltrials.gov (NCT02305940).

CT-based deep multi-label learning prediction model for outcome in patients with oropharyngeal squamous cell carcinoma.

BACKGROUND: Personalized treatment is increasingly required for oropharyngeal squamous cell carcinoma (OPSCC) patients due to emerging new cancer subtypes and treatment options. Outcome prediction model can help identify low or high-risk patients who may be suitable to receive de-escalation or intensified treatment approaches. PURPOSE: To develop a deep learning (DL)-based model for predicting multiple and associated efficacy endpoints in OPSCC patients based on computed tomography (CT). METHODS: Two patient cohorts were used in this study: a development cohort consisting of 524 OPSCC patients (70% for training and 30% for independent testing) and an external test cohort of 396 patients. Pre-treatment CT-scans with the gross primary tumor volume contours (GTVt) and clinical parameters were available to predict endpoints, including 2-year local control (LC), regional control (RC), locoregional control (LRC), distant metastasis-free survival (DMFS), disease-specific survival (DSS), overall survival (OS), and disease-free survival (DFS). We proposed DL outcome prediction models with the multi-label learning (MLL) strategy that integrates the associations of different endpoints based on clinical factors and CT-scans. RESULTS: The multi-label learning models outperformed the models that were developed based on a single endpoint for all endpoints especially with high AUCs ≥ 0.80 for 2-year RC, DMFS, DSS, OS, and DFS in the internal independent test set and for all endpoints except 2-year LRC in the external test set. Furthermore, with the models developed, patients could be stratified into high and low-risk groups that were significantly different for all endpoints in the internal test set and for all endpoints except DMFS in the external test set. CONCLUSION: MLL models demonstrated better discriminative ability for all 2-year efficacy endpoints than single outcome models in the internal test and for all endpoints except LRC in the external set.

Reverse Engineering of Digital Measures: Inviting Patients to the Conversation.

Background: Digital measures offer an unparalleled opportunity to create a more holistic picture of how people who are patients behave in their real-world environments, thereby establishing a better connection between patients, caregivers, and the clinical evidence used to drive drug development and disease management. Reaching this vision will require achieving a new level of co-creation between the stakeholders who design, develop, use, and make decisions using evidence from digital measures. Summary: In September 2022, the second in a series of meetings hosted by the Swiss Federal Institute of Technology in Zürich, the Foundation for the National Institutes of Health Biomarkers Consortium, and sponsored by Wellcome Trust, entitled "Reverse Engineering of Digital Measures," was held in Zurich, Switzerland, with a broad range of stakeholders sharing their experience across four case studies to examine how patient centricity is essential in shaping development and validation of digital evidence generation tools. Key Messages: In this paper, we discuss progress and the remaining barriers to widespread use of digital measures for evidence generation in clinical development and care delivery. We also present key discussion points and takeaways in order to continue discourse and provide a basis for dissemination and outreach to the wider community and other stakeholders. The work presented here shows us a blueprint for how and why the patient voice can be thoughtfully integrated into digital measure development and that continued multistakeholder engagement is critical for further progress.

Embracing complexity in sepsis.

Sepsis involves the dynamic interplay between a pathogen, the host response, the failure of organ systems, medical interventions and a myriad of other factors. This together results in a complex, dynamic and dysregulated state that has remained ungovernable thus far. While it is generally accepted that sepsis is very complex indeed, the concepts, approaches and methods that are necessary to understand this complexity remain underappreciated. In this perspective we view sepsis through the lens of complexity theory. We describe the concepts that support viewing sepsis as a state of a highly complex, non-linear and spatio-dynamic system. We argue that methods from the field of complex systems are pivotal for a fuller understanding of sepsis, and we highlight the progress that has been made over the last decades in this respect. Still, despite these considerable advancements, methods like computational modelling and network-based analyses continue to fly under the general scientific radar. We discuss what barriers contribute to this disconnect, and what we can do to embrace complexity with regards to measurements, research approaches and clinical applications. Specifically, we advocate a focus on longitudinal, more continuous biological data collection in sepsis. Understanding the complexity of sepsis will require a huge multidisciplinary effort, in which computational approaches derived from complex systems science must be supported by, and integrated with, biological data. Such integration could finetune computational models, guide validation experiments, and identify key pathways that could be targeted to modulate the system to the benefit of the host. We offer an example for immunological predictive modelling, which may inform agile trials that could be adjusted throughout the trajectory of disease. Overall, we argue that we should expand our current mental frameworks of sepsis, and embrace nonlinear, system-based thinking in order to move the field forward.