Diagnosis and Severity Assessment of COPD Using a Novel Fast-Response Capnometer and Interpretable Machine Learning.

INTRODUCTION: Spirometry is the gold standard for COPD diagnosis and severity determination, but is technique-dependent, nonspecific, and requires administration by a trained healthcare professional. There is a need for a fast, reliable, and precise alternative diagnostic test. This study's aim was to use interpretable machine learning to diagnose COPD and assess severity using 75-second carbon dioxide (CO2) breath records captured with TidalSense's N-TidalTM capnometer. METHOD: For COPD diagnosis, machine learning algorithms were trained and evaluated on 294 COPD (including GOLD stages 1-4) and 705 non-COPD participants. A logistic regression model was also trained to distinguish GOLD 1 from GOLD 4 COPD with the output probability used as an index of severity. RESULTS: The best diagnostic model achieved an AUROC of 0.890, sensitivity of 0.771, specificity of 0.850 and positive predictive value (PPV) of 0.834. Evaluating performance on all test capnograms that were confidently ruled in or out yielded PPV of 0.930 and NPV of 0.890. The severity determination model yielded an AUROC of 0.980, sensitivity of 0.958, specificity of 0.961 and PPV of 0.958 in distinguishing GOLD 1 from GOLD 4. Output probabilities from the severity determination model produced a correlation of 0.71 with percentage predicted FEV1. CONCLUSION: The N-TidalTM device could be used alongside interpretable machine learning as an accurate, point-of-care diagnostic test for COPD, particularly in primary care as a rapid rule-in or rule-out test. N-TidalTM also could be effective in monitoring disease progression, providing a possible alternative to spirometry for disease monitoring.

A scalable federated learning solution for secondary care using low-cost microcomputing: privacy-preserving development and evaluation of a COVID-19 screening test in UK hospitals.

BACKGROUND: Multicentre training could reduce biases in medical artificial intelligence (AI); however, ethical, legal, and technical considerations can constrain the ability of hospitals to share data. Federated learning enables institutions to participate in algorithm development while retaining custody of their data but uptake in hospitals has been limited, possibly as deployment requires specialist software and technical expertise at each site. We previously developed an artificial intelligence-driven screening test for COVID-19 in emergency departments, known as CURIAL-Lab, which uses vital signs and blood tests that are routinely available within 1 h of a patient's arrival. Here we aimed to federate our COVID-19 screening test by developing an easy-to-use embedded system-which we introduce as full-stack federated learning-to train and evaluate machine learning models across four UK hospital groups without centralising patient data. METHODS: We supplied a Raspberry Pi 4 Model B preloaded with our federated learning software pipeline to four National Health Service (NHS) hospital groups in the UK: Oxford University Hospitals NHS Foundation Trust (OUH; through the locally linked research University, University of Oxford), University Hospitals Birmingham NHS Foundation Trust (UHB), Bedfordshire Hospitals NHS Foundation Trust (BH), and Portsmouth Hospitals University NHS Trust (PUH). OUH, PUH, and UHB participated in federated training, training a deep neural network and logistic regressor over 150 rounds to form and calibrate a global model to predict COVID-19 status, using clinical data from patients admitted before the pandemic (COVID-19-negative) and testing positive for COVID-19 during the first wave of the pandemic. We conducted a federated evaluation of the global model for admissions during the second wave of the pandemic at OUH, PUH, and externally at BH. For OUH and PUH, we additionally performed local fine-tuning of the global model using the sites' individual training data, forming a site-tuned model, and evaluated the resultant model for admissions during the second wave of the pandemic. This study included data collected between Dec 1, 2018, and March 1, 2021; the exact date ranges used varied by site. The primary outcome was overall model performance, measured as the area under the receiver operating characteristic curve (AUROC). Removable micro secure digital (microSD) storage was destroyed on study completion. FINDINGS: Clinical data from 130 941 patients (1772 COVID-19-positive), routinely collected across three hospital groups (OUH, PUH, and UHB), were included in federated training. The evaluation step included data from 32 986 patients (3549 COVID-19-positive) attending OUH, PUH, or BH during the second wave of the pandemic. Federated training of a global deep neural network classifier improved upon performance of models trained locally in terms of AUROC by a mean of 27·6% (SD 2·2): AUROC increased from 0·574 (95% CI 0·560-0·589) at OUH and 0·622 (0·608-0·637) at PUH using the locally trained models to 0·872 (0·862-0·882) at OUH and 0·876 (0·865-0·886) at PUH using the federated global model. Performance improvement was smaller for a logistic regression model, with a mean increase in AUROC of 13·9% (0·5%). During federated external evaluation at BH, AUROC for the global deep neural network model was 0·917 (0·893-0·942), with 89·7% sensitivity (83·6-93·6) and 76·6% specificity (73·9-79·1). Site-specific tuning of the global model did not significantly improve performance (change in AUROC <0·01). INTERPRETATION: We developed an embedded system for federated learning, using microcomputing to optimise for ease of deployment. We deployed full-stack federated learning across four UK hospital groups to develop a COVID-19 screening test without centralising patient data. Federation improved model performance, and the resultant global models were generalisable. Full-stack federated learning could enable hospitals to contribute to AI development at low cost and without specialist technical expertise at each site. FUNDING: The Wellcome Trust, University of Oxford Medical and Life Sciences Translational Fund.

Randomized comparison of chest pain evaluation with FFRCT or standard care: Factors determining US costs.

BACKGROUND: FFRCT assesses the functional significance of lesions seen on CTCA, and may be a more efficient approach to chest pain evaluation. The FORECAST randomized trial found no significant difference in costs within the UK National Health Service, but implications for US costs are unknown. The purpose of this study was to compare costs in the FORECAST trial based on US healthcare cost weights, and to evaluate factors affecting costs. METHODS: Patients with stable chest pain were randomized either to the experimental strategy (CTCA with selective FFRCT), or to standard clinical pathways. Pre-randomization, the treating clinician declared the planned initial test. The primary outcome was nine-month cardiovascular care costs. RESULTS: Planned initial tests were CTCA in 912 patients (65%), stress testing in 393 (28%), and invasive angiography in 94 (7%). Mean US costs did not differ overall between the experimental strategy and standard care (cost difference +7% (+$324), CI -12% to +26%, p â€‹= â€‹0.49). Costs were 4% lower with the experimental strategy in the planned invasive angiography stratum (p for interaction â€‹= â€‹0.66). Baseline factors independently associated with costs were older age (+43%), male sex (+55%), diabetes (+37%), hypertension (+61%), hyperlipidemia (+94%), prior angina (+24%), and planned invasive angiography (+160%). Post-randomization cost drivers were coronary revascularization (+348%), invasive angiography (267%), and number of tests (+35%). CONCLUSIONS: Initial evaluation of chest pain using CTCA with FFRCT had similar US costs as standard care pathways. Costs were increased by baseline coronary risk factors and planned invasive angiography, and post-randomization invasive procedures and the number of tests. Registration at ClinicalTrials.gov (NCT03187639).

Fractional flow reserve derived from computed tomography coronary angiography in the assessment and management of stable chest pain: the FORECAST randomized trial.

AIMS: Fractional flow reserve (FFRCT) using computed tomography coronary angiography (CTCA) determines both the presence of coronary artery disease and vessel-specific ischaemia. We tested whether an evaluation strategy based on FFRCT would improve economic and clinical outcomes compared with standard care. METHODS AND RESULTS: Overall, 1400 patients with stable chest pain in 11 centres were randomized to initial testing with CTCA with selective FFRCT (experimental group) or standard clinical care pathways (standard group). The primary endpoint was total cardiac costs at 9 months. Secondary endpoints were angina status, quality of life, major adverse cardiac and cerebrovascular events, and use of invasive coronary angiography. Randomized groups were similar at baseline. Most patients had an initial CTCA: 439 (63%) in the standard group vs. 674 (96%) in the experimental group, 254 of whom (38%) underwent FFRCT. Mean total cardiac costs were higher by £114 (+8%) in the experimental group, with a 95% confidence interval from -£112 (-8%) to +£337 (+23%), though the difference was not significant (P = 0.10). Major adverse cardiac and cerebrovascular events did not differ significantly (10.2% in the experimental group vs. 10.6% in the standard group) and angina and quality of life improved to a similar degree over follow-up in both randomized groups. Invasive angiography was reduced significantly in the experimental group (19% vs. 25%, P = 0.01). CONCLUSION: A strategy of CTCA with selective FFRCT in patients with stable angina did not differ significantly from standard clinical care pathways in cost or clinical outcomes, but did reduce the use of invasive coronary angiography.

Nocturnal temperature-controlled laminar airflow device for adults with severe allergic asthma: the LASER RCT.

BACKGROUND: Severe asthma exacerbations are costly to patients and the NHS, and occur frequently in severely allergic patients. OBJECTIVE: To ascertain whether or not nocturnal temperature-controlled laminar airflow (TLA) device usage over 12 months can reduce severe exacerbations and improve asthma control and quality of life compared with a placebo device, while being cost-effective and acceptable to adults with severe allergic asthma. DESIGN: A pragmatic, multicentre, randomised, double-blind, placebo-controlled, parallel-group, superiority trial with qualitative interviews. The trial included an internal pilot with qualitative focus groups. SETTING: Fourteen hospitals in the UK that manage patients with severe asthma. PARTICIPANTS: Adults (16-75 years) with severe, poorly controlled, exacerbation-prone asthma despite high-intensity treatment, and who are sensitised to a perennial indoor aeroallergen. INTERVENTION: Nocturnal, home-based TLA treatment using an Airsonett® (Airsonett AB, Ängelholm, Sweden) device. The comparator was a placebo device that was identical to the active device except that it did not deliver the laminar airflow. Participants were allocated 1 : 1 to TLA therapy or placebo, minimised by site, origin of case, baseline severe exacerbation frequency, maintenance oral corticosteroid use and pre-bronchodilator forced expiratory volume in 1 second. MAIN OUTCOME MEASURES: Primary outcome - frequency of severe asthma exacerbations occurring within the 12-month follow-up period, defined as worsening of asthma requiring systemic corticosteroids [≥ 30 mg of prednisolone or equivalent daily (or ≥ 50% increase in dose if on maintenance dose of ≥ 30 mg of prednisolone)] for ≥ 3 days. Secondary outcomes - changes in asthma control, lung function, asthma-specific and global quality of life for participants, adherence to the intervention, device acceptability, health-care resource use and cost-effectiveness. RESULTS: Between May 2014 and January 2016, 489 patients consented to participate in the trial, of whom 249 failed screening and 240 were randomised (n = 119 in the treatment group and n = 121 in the placebo group); all were analysed. In total, 202 participants (84%) reported use of the device for 9-12 months. Qualitative analyses showed high levels of acceptability. The mean [standard deviation (SD)] rate of severe exacerbations did not differ between groups [active 1.39 (1.57), placebo 1.48 (2.03); risk ratio 0.92, 95% CI 0.66 to 1.27; p = 0.616]. There were no significant differences in secondary outcomes for lung function, except for a reduction in mean daily peak expiratory flow [mean (SD) difference 14.7 l/minute (7.35 l/minute), 95% CI 0.32 to 29.1 l/minute; p = 0.045) for those in the active device group. There were no differences in asthma control or airway inflammation and no serious harms related to the device. No significant difference between the groups in quality-adjusted life-years gained over 1 year was observed. In addition, there was no difference in generic or disease-specific health-related quality of life overall, although statistically significant higher quality of life at month 6 was observed. Increases in quality of life were not sufficient to offset the annual costs associated with use of the TLA device. LIMITATIONS: Missing outcome data could have resulted in an underestimation of exacerbations and rendered the study inconclusive. CONCLUSIONS: Within the limits of the data, no consistent benefits of the active device were demonstrated, and the differences observed were not sufficient to make the device cost-effective. The types of patients who may benefit from the TLA device, and the reasons for large reductions in exacerbation frequency in severe asthma trials, which also incorporate other methods of recording exacerbations, need to be explored. TRIAL REGISTRATION: Current Controlled Trials ISRCTN46346208. FUNDING: This project was funded by the NIHR Health Technology Assessment programme and will be published in full in Health Technology Assessment; Vol. 23, No. 29. See the NIHR Journals Library website for further project information.

Allergies (along with viruses) are common triggers of asthma exacerbations or 'attacks', which can cause suffering and frequent visits to the general practitioner or hospital. A new machine known as a temperature-controlled laminar airflow device, which remains at the bedside and is switched on every night, filters out allergy particles in the air of a patient's breathing zone, allowing their lungs to rest in clean air overnight. We tested whether or not this machine could improve the lives of those with severe allergic asthma. We recruited 240 people across 14 centres that treat severe asthma across the UK; approximately half received the active device and the other half received a machine that looked exactly the same but did not remove the allergens (a 'placebo' machine). One in five participants was recruited using newer methods of social media such as Facebook (Facebook, Inc., Menlo Park, CA, USA) and Twitter (Twitter, Inc., San Francisco, CA, USA). Participants found the machine easy to use and to live with and there were no significant side effects. The number of attacks reduced a lot in both participants using the active device and those who used the placebo device ­ two participants in five did not suffer any attacks during the trial. However, there was no difference in the number of attacks between the two groups. This might have been because participants did not record everything that happened to them. There was no difference in measurements showing how well the lungs were working, nor in participants' quality of life after 1 year of participating in the trial. Those who were interviewed told us that the study visits and questionnaires could be burdensome, although it was helpful to think more about their asthma. An improvement was seen in one aspect of participants' breathing as well as in their quality of life after 6 months of using the machine, but these potential health benefits could not outweigh the cost of the machine.

A multicentre non-blinded randomised controlled trial to assess the impact of regular early specialist symptom control treatment on quality of life in malignant mesothelioma (RESPECT-MESO): study protocol for a randomised controlled trial.

BACKGROUND: Malignant pleural mesothelioma is an incurable cancer caused by exposure to asbestos. The United Kingdom has the highest death rate from mesothelioma in the world and this figure is increasing. Median survival is 8 to 12 months, and most patients have symptoms at diagnosis. The fittest patients may be offered chemotherapy with palliative intent. For patients not fit for systemic anticancer treatment, best supportive care remains the mainstay of management. A study from the United States examining advanced lung cancer showed that early specialist palliative care input improved patient health related quality of life and depression symptoms 12 weeks after diagnosis. While mesothelioma and advanced lung cancer share many symptoms and have a poor prognosis, oncology and palliative care services in the United Kingdom, and many other countries, vary considerably compared to the United States. The aim of this trial is to assess whether regular early symptom control treatment provided by palliative care specialists can improve health related quality of life in patients newly diagnosed with mesothelioma. METHODS: This multicentre study is an non-blinded, randomised controlled, parallel group trial. A total of 174 patients with a new diagnosis of malignant pleural mesothelioma will be minimised with a random element in a 1:1 ratio to receive either 4 weekly regular early specialist symptom control care, or standard care. The primary outcome is health related quality of life for patients at 12 weeks. Secondary outcomes include health related quality of life for patients at 24 weeks, carer health related quality of life at 12 and 24 weeks, patient and carer mood at 12 and 24 weeks, overall survival and analysis of healthcare utilisation and cost. DISCUSSION: Current practice in the United Kingdom is to involve specialist palliative care towards the final weeks or months of a life-limiting illness. This study aims to investigate whether early, regular specialist care input can result in significant health related quality of life gains for patients with mesothelioma and if this change in treatment model is cost-effective. The results will be widely applicable to many institutions and patients both in the United Kingdom and internationally. TRIAL REGISTRATION: Current controlled trials ISRCTN18955704. Date ISRCTN assigned: 31 January 2014.