Incorporating uncertainty in learning to defer algorithms for safe computer-aided diagnosis.

Deep neural networks are increasingly being used for computer-aided diagnosis, but erroneous diagnoses can be extremely costly for patients. We propose a learning to defer with uncertainty (LDU) algorithm which identifies patients for whom diagnostic uncertainty is high and defers them for evaluation by human experts. LDU was evaluated on the diagnosis of myocardial infarction (using discharge summaries), the diagnosis of any comorbidities (using structured data), and the diagnosis of pleural effusion and pneumothorax (using chest x-rays), and compared with 'learning to defer without uncertainty information' (LD) and 'direct triage by uncertainty' (DT) methods. LDU achieved the same F1 score as LD but deferred considerably fewer patients (e.g. 36% vs. 69% deferral rate for diagnosing pleural effusion with an F1 score of 0.96). Furthermore, even when many patients were assigned the wrong diagnosis with high confidence (e.g. for the diagnosis of any comorbidities) LDU achieved a 17% increase in F1 score, whereas DT was not applicable. Importantly, the weight of the defer loss in LDU can be easily adjusted to obtain the desired trade-off between diagnostic accuracy and deferral rate. In conclusion, LDU can readily augment any existing diagnostic network to reduce the risk of erroneous diagnoses in clinical practice.

Diagnostic accuracy of different computer-aided diagnostic systems for malignant and benign thyroid nodules classification in ultrasound images: A systematic review and meta-analysis protocol.

OBJECTIVE: The aim of this study was to determine the diagnostic accuracy of different computer-aided diagnostic (CAD) systems for thyroid nodules classification. METHODS: A systematic search of the literature was conducted from inception until March, 2019 using the PubMed, EMBASE, Web of science, and Cochrane library. Literature selection and data extraction were conducted by 2 independent reviewers. Numerical values for sensitivity and specificity were obtained from false negative (FN), false positive (FP), true negative (TN), and true positive (TP) rates, presented alongside graphical representations with boxes marking the values and horizontal lines showing the confidence intervals (CIs). Summary receiver operating characteristic (SROC) curves were applied to assess the performance of diagnostic tests. Data were processed using Review Manager 5.3 and Stata 15. The methodological quality of included studies was assessed using Quality Assessment of Diagnostic Accuracy Studies (QUADAS-2) tool. TRIAL REGISTRATION NUMBER: PROSPERO CRD42019132540.

An exploration of the ecological validity of the Virtual Action Planning-Supermarket (VAP-S) with people with schizophrenia.

People with schizophrenia often have functional limitations that affect their daily activities due to executive function deficits. One way to assess these deficits is through the use of virtual reality programmes that reproduce real-life instrumental activities of daily living (IADLs). One such programme is the Virtual Action Planning-Supermarket (VAP-S). This exploratory study aimed to examine the ecological validity of this programme, specifically, how task performance in both virtual and natural environments compares. Case studies were used and involved five participants with schizophrenia, who were familiar with grocery shopping. They were assessed during both the VAP-S shopping task and a real-life grocery shopping task using an observational assessment tool, the Perceive, Recall, Plan and Perform (PRPP) System of Task Analysis. The results show that when difficulties were present in the virtual task, difficulties were also observed in the real-life task. For some participants, greater difficulties were observed in the virtual task. These difficulties could be explained by the presence of perceptual deficits and problems remembering the required sequenced actions in the virtual task. In conclusion, performance on the VAP-S by these five participants was generally comparable to the performance in a natural environment.

[Optimisation of subcutaneous defibrillator programming after inappropriate shocks due to new onset of right bundle branch block]. / Optimierung der Geräteeinstellung eines subkutanen Defibrillators nach inadäquaten Schocks durch neu aufgetretenen Rechtsschenkelblock.

The subcutaneous implantable defibrillator (S-ICD) has become an established tool for the prevention of sudden cardiac death. Based on its detection properties, the S­ICD is essentially dependent on correct morphology discrimination of the QRS complex and avoidance of potential T­wave sensing. We report on a patient who experienced multiple inappropriate S­ICD shocks due to T­wave oversensing in the setting of new onset of right bundle branch block. Strategies for the optimisation of the device programming are discussed.

[Redetection of tachycardia in severe ventricular undersensing]. / Redetektion einer Tachykardie bei ausgeprägtem ventrikulären Undersensing.

In a 50-year-old patient with arrhythmogenic right ventricular cardiomyopathy (ARVC) and implantable cardioverter defibrillator (ICD) two shock discharges occurred after several ineffective attempts with antitachycardia pacing. The analysis of the stored electrograms shows a peculiarity of shocks with low energy, a problem of ICD therapy in ARVC, and the impact of committed shocks as opposed to non-committed shocks.

[Mismanagement in the arrhythmia recognition algorithm of the LifeVest® with consecutive fatal outcome]. / Missmanagement im Arrhythmieerkennungsalgorithmus der LifeVest® mit konsekutivem letalem Verlauf.

We describe for the first time the misinterpretation of a wearable cardioverter defibrillator in the arrhythmia recognition algorithm with subsequent fatal outcome of a multi-morbid patient with an ischemic cardiomyopathy and a highly reduced left ventricular pump function (30 %). The patient's death was preceded by a life-threatening shockable rhythm which was repeatedly documented, but ultimately not correctly recognized by the system and therefore not treated.

Pacemaker-mediated tachycardia initiated by an atrioventricular search algorithm to minimize right ventricular pacing.

We report the initiation of pacemaker-mediated tachycardia by a St Jude implantable cardioverter-defibrillator with a programmed Ventricular Intrinsic Preference algorithm used for minimizing or inhibiting right ventricular pacing. This feature prolongs the atrioventricular (AV) delay periodically to determine if ventricular sensed events follow atrial events. Retrograde ventriculoatrial conduction and pacemaker-mediated tachycardia were initiated by long extended AV delays of 300 and 400 milliseconds. The 400-millisecond AV delay consisted of the programmed sensed AV delay (100 milliseconds) plus the Ventricular Intrinsic Preference increment (200 milliseconds) plus 100 milliseconds imposed by the AutoCapture algorithm when it detected loss of ventricular capture.

Temporary morphological changes in plus disease induced during contact digital imaging.

OBJECTIVE: To compare and quantify the retinal vascular changes induced by non-intentional pressure contact by digital handheld camera during retinopathy of prematurity (ROP) imaging by means of a computer-based image analysis system, Retinal Image multiScale Analysis. METHODS: A set of 10 wide-angle retinal pairs of photographs per patient, who underwent routine ROP examinations, was measured. Vascular trees were matched between 'compression artifact' (absence of the vascular column at the optic nerve) and 'not compression artifact' conditions. Parameters were analyzed using a two-level linear model for each individual parameter for arterioles and venules separately: integrated curvature (IC), diameter (d), and tortuosity index (TI). RESULTS: Images affected with compression artifact showed significant vascular d (P<0.01) changes in both arteries and veins, as well as in artery IC (P<0.05). Vascular TI remained unchanged in both groups. CONCLUSIONS: Non-adverted corneal pressure with the RetCam lens could compress and decrease intra-arterial diameter or even collapse retinal vessels. Careful attention to technique is essential to avoid absence of the arterial blood column at the optic nerve head that is indicative of increased pressure during imaging.

A descriptive analysis of seizure events among adults who participated in a computer-based assessment.

The purpose of this study was to document seizure events associated with the use of a computer-based assessment and to describe the contextual factors surrounding these seizure episodes. Study participants were adults with epilepsy who were enrolled at research sites in Atlanta and Boston. Subjects were asked to complete a computer-based assessment at 3 time points. Fourteen seizure events were documented; they occurred during 1.6% of all completed assessments (896) and affected 4.4% of the participants (320). The mean age of participants who experienced seizure events was 41.4 years; about 70% were female, and 70% were white. A variety of possible precipitating factors for seizure events included hunger, fatigue, stress, and medication changes. Participants indicated computer use could have triggered their seizures in 2 instances. These findings suggest use of computer-based assessments may pose minimal risks for adults with epilepsy, particularly those without a history of photosensitivity epilepsy.

Medicina baseada em evidências: análise crítica / Medicine based on evidences: critical analysis

Os autores conceituam a medicina baseada em evidências e referenciam os critérios e as razões dos defensores desta estratégia. Objetivam mostrar que há riscos na sacralização deste novo paradigma assistencial e pedagógico na sua forma de alcançar a verdade absoluta, principalmente pelas dificuldades na aquisição de publicações de alto nível e no fato de que a Medicina clínica é uma arte e não uma ciência exata. Coloca-se em dúvida a absolutização da idéia de um padrão em saúde baseado em evidências e o risco decorrente de uma profusão alarmante de publicações, algumas em franca oposição com as outras, o que, na visão dos articulistas, complica ainda mais a situação do médico. Afirmam que o ideal seria buscar um equilíbrio entre as teorias fisiopatológicas consagradas, na experiência pessoal e no resultado das investigações clínicas

An objective analysis of process errors in trauma resuscitations.

OBJECTIVE: A computer-based system to apply trauma resuscitation protocols to patients with penetrating thoracoabdominal trauma was previously validated for 97 consecutive patients at a Level 1 trauma center by a panel of the trauma attendings and further refined by a panel of national trauma experts. The purpose of this article is to describe how this system is now used to objectively critique the actual care given to those patients for process errors in reasoning, independent of outcome. METHODS: A chronological narrative of the care of each patient was presented to the computer program. The actual care was compared with the validated computer protocols at each decision point and differences were classified by a predetermined scoring system from 0 to 100, based on the potential impact on outcome, as critical/noncritical/no errors of commission, omission, or procedure selection. RESULTS: Errors in reasoning occurred in 100% of the 97 cases studied, averaging 11.9/case. Errors of omission were more prevalent than errors of commission (2. 4 errors/case vs 1.2) and were of greater severity (19.4/error vs 5. 1). The largest number of errors involved the failure to record, and perhaps observe, beside information relevant to the reasoning process, an average of 7.4 missing items/patient. Only 2 of the 10 adverse outcomes were judged to be potentially related to errors of reasoning. CONCLUSIONS: Process errors in reasoning were ubiquitous, occurring in every case, although they were infrequently judged to be potentially related to an adverse outcome. Errors of omission were assessed to be more severe. The most common error was failure to consider, or document, available relevant information in the selection of appropriate care.

A new procedure for frameless computer navigated stereotaxy.

OBJECTIVE: Stereotactic procedures using frame-based systems have become well established in neurosurgery. Later, stereotactic computers have provided the neurosurgeon with a broader range of applications. A new, frameless stereotactic guide which utilizes the navigational abilities of a computer has been developed. This clinical study evaluates the accuracy and safety of the system when applied for puncture of tumours and abscesses in the cerebrum. METHODS: Using a frameless setup 36 patients were operated on for a total of 39 intracranial processes over a period of two years. Three patients were operated on twice. Computer data, time of surgery, anesthesia, complications, tumour localization and tissue examinations were recorded. RESULTS: Biopsies were obtained from all lobes of the cerebrum and a diagnosis established in all cases except two. Mean age and total range was 52 years and 15-82 years, respectively. Median time of surgery was approximately 60 minutes including positioning of the patient and the registration process on the stereotactic computer. Local anesthesia was used for 25 operations and general anesthesia for 14 operations. There was one postoperative abscess formation in a previously immunosuppressed patient and a per-operative epileptic seizure caused by electrocoagulation of the dura in a second patient. CONCLUSIONS: Based on this clinical study a new procedure for obtaining stereotactic biopsies of intracranial processes is introduced. All tumours were successfully reached with two cases of complications.