Collaboration between explainable artificial intelligence and pulmonologists improves the accuracy of pulmonary function test interpretation.

BACKGROUND: Few studies have investigated the collaborative potential between artificial intelligence (AI) and pulmonologists for diagnosing pulmonary disease. We hypothesised that the collaboration between a pulmonologist and AI with explanations (explainable AI (XAI)) is superior in diagnostic interpretation of pulmonary function tests (PFTs) than the pulmonologist without support. METHODS: The study was conducted in two phases, a monocentre study (phase 1) and a multicentre intervention study (phase 2). Each phase utilised two different sets of 24 PFT reports of patients with a clinically validated gold standard diagnosis. Each PFT was interpreted without (control) and with XAI's suggestions (intervention). Pulmonologists provided a differential diagnosis consisting of a preferential diagnosis and optionally up to three additional diagnoses. The primary end-point compared accuracy of preferential and additional diagnoses between control and intervention. Secondary end-points were the number of diagnoses in differential diagnosis, diagnostic confidence and inter-rater agreement. We also analysed how XAI influenced pulmonologists' decisions. RESULTS: In phase 1 (n=16 pulmonologists), mean preferential and differential diagnostic accuracy significantly increased by 10.4% and 9.4%, respectively, between control and intervention (p<0.001). Improvements were somewhat lower but highly significant (p<0.0001) in phase 2 (5.4% and 8.7%, respectively; n=62 pulmonologists). In both phases, the number of diagnoses in the differential diagnosis did not reduce, but diagnostic confidence and inter-rater agreement significantly increased during intervention. Pulmonologists updated their decisions with XAI's feedback and consistently improved their baseline performance if AI provided correct predictions. CONCLUSION: A collaboration between a pulmonologist and XAI is better at interpreting PFTs than individual pulmonologists reading without XAI support or XAI alone.

Correlation of symptoms and physical activity level in chronic obstructive pulmonary disease patients: results from the observational SPACE study.

OBJECTIVES: Chronic obstructive pulmonary disease (COPD) is characterized by persistent airflow limitation and high symptom burden that interferes with physical activity and results in a vicious cycle of inactivity and symptom worsening. The aim of this multicenter, observational study was to determine the prevalence and severity of morning, daytime and night-time symptoms as well as patterns of physical activity levels (PALs) and their interrelation in Belgian COPD patients, enrolled in the multinational SPACE study (NCT03031769). METHODS: Socio-demographic, socio-economic and disease characteristics data were collected from patients' medical records as part of a routine visit to their primary care practice or pulmonologist. Dedicated questionnaires were used to evaluate respiratory symptoms for each part of the day. PAL was assessed by means of self- and interview-reported tools, and physician's judgment. Patients were also classified according to GOLD recommendations 2013 and 2017. RESULTS: Overall, 102 Belgian patients participated in the study (mean age 67 years, 60.8% males). Over 85% of patients experienced respiratory symptoms throughout the day and about one-third were considered as 'active' (PAL ≥150 minutes/week). Physician-assessed PALs were higher than self-reported PALs, categorizing fewer patients as 'inactive' (17.6% versus 42.2%, respectively). PALs and symptoms were weakly interrelated. Inactive patients were present in all GOLD classification groups. CONCLUSION: Stable Belgian COPD patients enrolled in the SPACE study presented 24-hour respiratory symptoms and insufficient PALs. Physicians tended to overestimate patients' physical activity. Inactive patients were present across all GOLD classification groups. New approaches are deemed necessary to objectively identify and activate sedentary patients.

Treatment failure and hospital readmissions in severe COPD exacerbations treated with azithromycin versus placebo - a post-hoc analysis of the BACE randomized controlled trial.

BACKGROUND: In the BACE trial, a 3-month (3 m) intervention with azithromycin, initiated at the onset of an infectious COPD exacerbation requiring hospitalization, decreased the rate of a first treatment failure (TF); the composite of treatment intensification (TI), step-up in hospital care (SH) and mortality. OBJECTIVES: (1) To investigate the intervention's effect on recurrent events, and (2) to identify clinical subgroups most likely to benefit, determined from the incidence rate of TF and hospital readmissions. METHODS: Enrolment criteria included the diagnosis of COPD, a smoking history of ≥10 pack-years and ≥ 1 exacerbation in the previous year. Rate ratio (RR) calculations, subgroup analyses and modelling of continuous variables using splines were based on a Poisson regression model, adjusted for exposure time. RESULTS: Azithromycin significantly reduced TF by 24% within 3 m (RR = 0.76, 95%CI:0.59;0.97, p = 0.031) through a 50% reduction in SH (RR = 0.50, 95%CI:0.30;0.81, p = 0.006), which comprised of a 53% reduction in hospital readmissions (RR = 0.47, 95%CI:0.27;0.80; p = 0.007). A significant interaction between the intervention, CRP and blood eosinophil count at hospital admission was found, with azithromycin significantly reducing hospital readmissions in patients with high CRP (> 50 mg/L, RR = 0.18, 95%CI:0.05;0.60, p = 0.005), or low blood eosinophil count (<300cells/µL, RR = 0.33, 95%CI:0.17;0.64, p = 0.001). No differences were observed in treatment response by age, FEV1, CRP or blood eosinophil count in continuous analyses. CONCLUSIONS: This post-hoc analysis of the BACE trial shows that azithromycin initiated at the onset of an infectious COPD exacerbation requiring hospitalization reduces the incidence rate of TF within 3 m by preventing hospital readmissions. In patients with high CRP or low blood eosinophil count at admission this treatment effect was more pronounced, suggesting a potential role for these biomarkers in guiding azithromycin therapy. TRIAL REGISTRATION: ClinicalTrials.gov number. NCT02135354 .

Azithromycin during Acute Chronic Obstructive Pulmonary Disease Exacerbations Requiring Hospitalization (BACE). A Multicenter, Randomized, Double-Blind, Placebo-controlled Trial.

Rationale: Azithromycin prevents acute exacerbations of chronic obstructive pulmonary disease (AECOPDs); however, its value in the treatment of an AECOPD requiring hospitalization remains to be defined.Objectives: We investigated whether a 3-month intervention with low-dose azithromycin could decrease treatment failure (TF) when initiated at hospital admission and added to standard care.Methods: In an investigator-initiated, multicenter, randomized, double-blind, placebo-controlled trial, patients who had been hospitalized for an AECOPD and had a smoking history of ≥10 pack-years and one or more exacerbations in the previous year were randomized (1:1) within 48 hours of hospital admission to azithromycin or placebo. The study drug (500 mg/d for 3 d) was administered on top of a standardized acute treatment of systemic corticosteroids and antibiotics, and subsequently continued for 3 months (250 mg/2 d). The patients were followed for 6 months thereafter. Time-to-first-event analyses evaluated the TF rate within 3 months as a novel primary endpoint in the intention-to-treat population, with TF defined as the composite of treatment intensification with systemic corticosteroids and/or antibiotics, a step-up in hospital care or readmission for respiratory reasons, or all-cause mortality.Measurements and Main Results: A total of 301 patients were randomized to azithromycin (n = 147) or placebo (n = 154). The TF rate within 3 months was 49% in the azithromycin group and 60% in the placebo group (hazard ratio, 0.73; 95% confidence interval, 0.53-1.01; P = 0.0526). Treatment intensification, step-up in hospital care, and mortality rates within 3 months were 47% versus 60% (P = 0.0272), 13% versus 28% (P = 0.0024), and 2% versus 4% (P = 0.5075) in the azithromycin and placebo groups, respectively. Clinical benefits were lost 6 months after withdrawal.Conclusions: Three months of azithromycin for an infectious AECOPD requiring hospitalization may significantly reduce TF during the highest-risk period. Prolonged treatment seems to be necessary to maintain clinical benefits.

Artificial intelligence outperforms pulmonologists in the interpretation of pulmonary function tests.

The interpretation of pulmonary function tests (PFTs) to diagnose respiratory diseases is built on expert opinion that relies on the recognition of patterns and the clinical context for detection of specific diseases. In this study, we aimed to explore the accuracy and interrater variability of pulmonologists when interpreting PFTs compared with artificial intelligence (AI)-based software that was developed and validated in more than 1500 historical patient cases.120 pulmonologists from 16 European hospitals evaluated 50 cases with PFT and clinical information, resulting in 6000 independent interpretations. The AI software examined the same data. American Thoracic Society/European Respiratory Society guidelines were used as the gold standard for PFT pattern interpretation. The gold standard for diagnosis was derived from clinical history, PFT and all additional tests.The pattern recognition of PFTs by pulmonologists (senior 73%, junior 27%) matched the guidelines in 74.4±5.9% of the cases (range 56-88%). The interrater variability of κ=0.67 pointed to a common agreement. Pulmonologists made correct diagnoses in 44.6±8.7% of the cases (range 24-62%) with a large interrater variability (κ=0.35). The AI-based software perfectly matched the PFT pattern interpretations (100%) and assigned a correct diagnosis in 82% of all cases (p<0.0001 for both measures).The interpretation of PFTs by pulmonologists leads to marked variations and errors. AI-based software provides more accurate interpretations and may serve as a powerful decision support tool to improve clinical practice.

Paraneoplastic Cushing syndrome, case-series and review of the literature.

OBJECTIVES: Paraneoplastic Cushing syndrome is a rare condition, representing a small fraction of the adrenocorticotropic hormone (ACTH)-dependent cases of Cushing syndrome Methods: Four case descriptions and literature review, highlighting the diagnostic challenges and treatment options are presented. RESULTS: Different tumor types can be associated with ectopic ACTH secretion. The most common types are bronchial carcinoids and small cell lung carcinoma (SCLC). However, in approximately 10 to 20% of the cases, no overt tumor (occult tumor) can be found. The diagnosis is made in a multistep process. Firstly, hypercortisolemia and adrenocorticotropin hormone dependency have to be confirmed. Distinction between a pituitary or ectopic cause can be cumbersome. MRI of the pituitary gland, a corticotropin releasing hormone stimulation test and a sinus petrosus sampling can be used. Treatment options consist of tumor management, somatostatin analogs, steroidogenesis inhibitors, and bilateral adrenalectomy. CONCLUSION: Clinicians should consider the diagnosis, and opt for specific treatment, especially in patients with a history of neuroendocrine tumors.

Rehabilitation in patients with radically treated respiratory cancer: A randomised controlled trial comparing two training modalities.

INTRODUCTION: The evidence on the effectiveness of rehabilitation in lung cancer patients is limited. Whole body vibration (WBV) has been proposed as an alternative to conventional resistance training (CRT). METHODS: We investigated the effect of radical treatment (RT) and of two rehabilitation programmes in lung cancer patients. The primary endpoint was a change in 6-min walking distance (6MWD) after rehabilitation. Patients were randomised after RT to either CRT, WBVT or standard follow-up (CON). Patients were evaluated before, after RT and after 12 weeks of intervention. RESULTS: Of 121 included patients, 70 were randomised to either CON (24), CRT (24) or WBVT (22). After RT, 6MWD decreased with a mean of 38m (95% CI 22-54) and increased with a mean of 95m (95% CI 58-132) in CRT (p<0.0001), 37m (95% CI -1-76) in WBVT (p=0.06) and 1m (95% CI -34-36) in CON (p=0.95), respectively. Surgical treatment, magnitude of decrease in 6MWD by RT and allocation to either CRT or WBVT were prognostic for reaching the minimally clinically important difference of 54m increase in 6MWD after intervention. CONCLUSIONS: RT of lung cancer significantly impairs patients' exercise capacity. CRT significantly improves and restores functional exercise capacity, whereas WBVT does not fully substitute for CRT.

Renal angiomyolipoma in association with pulmonary lymphangioleiomyomatosis.

The authors report on a premenopausal female hemodialysis patient with relapsing pneumothorax, in whom the diagnosis of pulmonary lymphangioleiomyomatosis (LAM) was made. Ten years earlier, she had retroperitoneal bleeding from a kidney tumor corresponding to an angiomyolipoma (AML). The association between renal AML and pulmonary LAM is reviewed. Renal AML represents the most frequent extrapulmonary manifestation of pulmonary LAM. It is found in 32% to 60 % of cases in which a systematic search with abdominal computed tomography (CT) scan is done. The latter approach is advised to help avoid complications caused by renal AML. Therapeutic recommendations for renal AML are based on tumor size or presence of symptoms. Conversely, premenopausal women presenting with AML should be investigated for associated pulmonary LAM with high-resolution CT scan. Because LAM is very likely estrogen dependent, one of the several proposed antiestrogen therapies should be considered. Finally, there is significant overlap between renal AML, pulmonary LAM, and tuberous sclerosis. The latter should therefore be actively searched for in case of either AML or LAM.