Deep Learning Based Long Term Mortality Prediction in the National Lung Screening Trial

In this study, the long-term mortality in the National Lung Screening Trial (NLST) was investigated using a deep learning-based method. Binary classification of the non-lung-cancer mortality (i.e. cardiovascular and respiratory mortality) was performed using neural network models centered around a 3D-ResNet. The models were trained on a participant age, gender, and smoking history matched cohort. Utilising both the 3D CT scan and clinical information, the models can achieve an AUC of 0.73 which outperforms humans at cardiovascular mortality prediction. By interpreting the trained models with 3D saliency maps, we examined the features on the CT scans that correspond to the mortality signal. The saliency maps can potentially assist the clinicians' and radiologists' to identify regions of concern on the image that may indicate the need to adopt preventative healthcare management strategies to prolong the patients' life expectancy.

Creating a training set for artificial intelligence from initial segmentations of airways

Airways segmentation is important for research about pulmonary disease but require a large amount of time by trained specialists. We used an openly available software to improve airways segmentations obtained from an artificial intelligence (AI) tool and retrained the tool to get a better performance. Fifteen initial airway segmentations from low-dose chest computed tomography scans were obtained with a 3D-Unet AI tool previously trained on Danish Lung Cancer Screening Trial and Erasmus-MC Sophia datasets. Segmentations were manually corrected in 3D Slicer. The corrected airway segmentations were used to retrain the 3D-Unet. Airway measurements were automatically obtained and included count, airway length and luminal diameter per generation from the segmentations. Correcting segmentations required 2–4 h per scan. Manually corrected segmentations had more branches (p < 0.001), longer airways (p < 0.001) and smaller luminal diameters (p = 0.004) than initial segmentations. Segmentations from retrained 3D-Unets trended towards more branches and longer airways compared to the initial segmentations. The largest changes were seen in airways from 6th generation onwards. Manual correction results in significantly improved segmentations and is potentially a useful and time-efficient method to improve the AI tool performance on a specific hospital or research dataset.

Improved User Engagement and Assessment of Treatment Effectiveness in Patients Utilizing a Novel Digital Mobile Health Application During Spinal Cord Stimulation Screening Trials (Preprint)

BACKGROUND Patient outcomes and experience during a Spinal Cord Stimulation (SCS) screening trial can have a significant effect on whether or not to proceed with long-term, permanent implantation of an SCS device for treatment of chronic pain. Enhancing the ability to track and assess patients during this initial trial evaluation offers the potential for improved understanding regarding the suitability of permanent device implantation as well as identification of the particular SCS-based neurostimulative modalities and/or parameters that may provide substantial analgesia in a patient-specific manner. OBJECTIVE In this report, we describe a preliminary, real-world assessment of a new, real-time tracking, smart device-based digital application (app) used by patients with chronic pain undergoing trial screening for SCS therapy. METHODS This is a real-world, retrospective evaluation of 13,331 patients diagnosed with chronic pain who utilized the new “mySCS” mobile application (Boston Scientific, Valencia, CA) during an SCS screening trial. The app design is HIPAA-compliant and compatible with most commercially available smartphones (e.g., Apple ® iPhone ®, Android ®). The app enables tracking of user-inputted health-related responses (i.e., pain relief, activity level, and/or sleep quality) in addition to personal trial goals and a summary of overall experience during the SCS trial. A de-identified, aggregate analysis of user engagement, user-submitted responses, and overall trial success was conducted. RESULTS When provided the opportunity, the percentage of users who engaged with the tracking app for ≥50% of the time during their trial was found to be 64.5%. Among those who used the app, ~58% (n = 7795) entered a trial goal(s). Most patients underwent SCS screening with a trial duration of at least 7-days (n = 7739). Of those patients who undertook a 7-day SCS trial, 62.3% engaged the app for 4-days or more. In addition, among all who submitted descriptive responses using the app, health-related improvements were reported by 78% of patients who reached day 3 of the screening phase assessment and by 83% of those who reached trial completion. A trial success rate of 91% was determined for those who used the app (versus 85% success rate for non-users). CONCLUSIONS Data from this initial, real-world examination of a mobile, digital-health-based tracking application (“mySCS” app), as utilized during the SCS screening phase, demonstrates that substantial patient engagement can be achieved while also providing for the acquisition of more real-time and direct from patient outcome measures that may help facilitate improved SCS trial success.