Cancer treatments are often more successful when the disease is detected early. We evaluated the feasibility and safety of multicancer blood testing coupled with positron emission tomography-computed ...tomography (PET-CT) imaging to detect cancer in a prospective, interventional study of 10,006 women not previously known to have cancer. Positive blood tests were independently confirmed by a diagnostic PET-CT, which also localized the cancer. Twenty-six cancers were detected by blood testing. Of these, 15 underwent PET-CT imaging and nine (60%) were surgically excised. Twenty-four additional cancers were detected by standard-of-care screening and 46 by neither approach. One percent of participants underwent PET-CT imaging based on false-positive blood tests, and 0.22% underwent a futile invasive diagnostic procedure. These data demonstrate that multicancer blood testing combined with PET-CT can be safely incorporated into routine clinical care, in some cases leading to surgery with intent to cure.
The electrocardiogram (ECG) is a widely used medical test, consisting of voltage versus time traces collected from surface recordings over the heart
. Here we hypothesized that a deep neural network ...(DNN) can predict an important future clinical event, 1-year all-cause mortality, from ECG voltage-time traces. By using ECGs collected over a 34-year period in a large regional health system, we trained a DNN with 1,169,662 12-lead resting ECGs obtained from 253,397 patients, in which 99,371 events occurred. The model achieved an area under the curve (AUC) of 0.88 on a held-out test set of 168,914 patients, in which 14,207 events occurred. Even within the large subset of patients (n = 45,285) with ECGs interpreted as 'normal' by a physician, the performance of the model in predicting 1-year mortality remained high (AUC = 0.85). A blinded survey of cardiologists demonstrated that many of the discriminating features of these normal ECGs were not apparent to expert reviewers. Finally, a Cox proportional-hazard model revealed a hazard ratio of 9.5 (P < 0.005) for the two predicted groups (dead versus alive 1 year after ECG) over a 25-year follow-up period. These results show that deep learning can add substantial prognostic information to the interpretation of 12-lead resting ECGs, even in cases that are interpreted as normal by physicians.
The COVID-19 pandemic is stretching medical resources internationally, sometimes creating ventilator shortages that complicate clinical and ethical situations. The possibility of needing to ventilate ...multiple patients with a single ventilator raises patient health and safety concerns in addition to clinical conditions needing treatment. Wherever ventilators are employed, additional tubing and splitting adaptors may be available. Adjustable flow-compensating resistance for differences in lung compliance on individual limbs may not be readily implementable. By exploring a number and range of possible contributing factors using computational simulation without risk of patient harm, this paper attempts to define useful bounds for ventilation parameters when compensatory resistance in limbs of a shared breathing circuit is not possible. This desperate approach to shared ventilation support would be a last resort when alternatives have been exhausted.
A whole-body computational physiology model (using lumped parameters) was used to simulate each patient being ventilated. The primary model of a single patient with a dedicated ventilator was augmented to model two patients sharing a single ventilator. In addition to lung mechanics or estimation of CO2 and pH expected for set ventilation parameters (considerations of lung physiology alone), full physiological simulation provides estimates of additional values for oxyhemoglobin saturation, arterial oxygen tension, and other patient parameters. A range of ventilator settings and patient characteristics were simulated for paired patients.
To be useful for clinicians, attention has been directed to clinically available parameters. These simulations show patient outcome during multi-patient ventilation is most closely correlated to lung compliance, oxygenation index, oxygen saturation index, and end-tidal carbon dioxide of individual patients. The simulated patient outcome metrics were satisfactory when the lung compliance difference between two patients was less than 12 mL/cmH2O, and the oxygen saturation index difference was less than 2 mmHg.
In resource-limited regions of the world, the COVID-19 pandemic will result in equipment shortages. While single-patient ventilation is preferable, if that option is unavailable and ventilator sharing using limbs without flow resistance compensation is the only available alternative, these simulations provide a conceptual framework and guidelines for clinical patient selection.
Intracranial hemorrhage (ICH) requires prompt diagnosis to optimize patient outcomes. We hypothesized that machine learning algorithms could automatically analyze computed tomography (CT) of the ...head, prioritize radiology worklists and reduce time to diagnosis of ICH. 46,583 head CTs (~2 million images) acquired from 2007-2017 were collected from several facilities across Geisinger. A deep convolutional neural network was trained on 37,074 studies and subsequently evaluated on 9499 unseen studies. The predictive model was implemented prospectively for 3 months to re-prioritize "routine" head CT studies as "stat" on realtime radiology worklists if an ICH was detected. Time to diagnosis was compared between the re-prioritized "stat" and "routine" studies. A neuroradiologist blinded to the study reviewed false positive studies to determine whether the dictating radiologist overlooked ICH. The model achieved an area under the ROC curve of 0.846 (0.837-0.856). During implementation, 94 of 347 "routine" studies were re-prioritized to "stat", and 60/94 had ICH identified by the radiologist. Five new cases of ICH were identified, and median time to diagnosis was significantly reduced (
< 0.0001) from 512 to 19 min. In particular, one outpatient with vague symptoms on anti-coagulation was found to have an ICH which was treated promptly with reversal of anticoagulation, resulting in a good clinical outcome. Of the 34 false positives, the blinded over-reader identified four probable ICH cases overlooked in original interpretation. In conclusion, an artificial intelligence algorithm can prioritize radiology worklists to reduce time to diagnosis of new outpatient ICH by 96% and may also identify subtle ICH overlooked by radiologists. This demonstrates the positive impact of advanced machine learning in radiology workflow optimization.
The global population is becoming more overweight and obese, leading to increases in associated morbidity and mortality rates. Advances in catheter-directed embolotherapy offer the potential for the ...interventional radiologist to make a contribution to weight loss. Left gastric artery embolization reduces the supply of blood to the gastric fundus and decreases serum levels of ghrelin. Early evidence suggests that this alteration in gut hormone balance leads to changes in energy homeostasis and weight reduction. The pathophysiologic findings and current evidence associated with the use of left gastric artery embolization are reviewed.
The prevalence of obesity continues to increase at an alarming rate, and, thus far, advances in medical management have been relatively ineffective in slowing this trend. Lifestyle modifications such as diet and exercise are effective initially, but most patients regain the weight in the long term. Bariatric surgery is the most effective strategy for achieving long-term weight loss; however, as with all surgical procedures, it has potential complications.
With the increased availability of three‐dimensional (3D) printers, innovative teaching and training materials have been created in medical fields. For pathology, the use of 3D printing has been ...largely limited to anatomic representations of disease processes or the development of supplies during the coronavirus disease 2019 pandemic. Herein, an institution's 3D printing laboratory and staff with expertise in additive manufacturing illustrate how this can address design issues in cytopathology specimen collection and processing. The authors' institutional 3D printing laboratory, along with students and trainees, used computer‐aided design and 3D printers to iterate on design, create prototypes, and generate final usable materials using additive manufacturing. The program Microsoft Forms was used to solicit qualitative and quantitative feedback. The 3D‐printed models were created to assist with cytopreparation, rapid on‐site evaluation, and storage of materials in the preanalytical phase of processing. These parts provided better organization of materials for cytology specimen collection and staining, in addition to optimizing storage of specimens with multiple sized containers to optimize patient safety. The apparatus also allowed liquids to be stabilized in transport and removed faster at the time of rapid on‐site evaluation. Rectangular boxes were also created to optimally organize all components of a specimen in cytopreparation to simplify and expedite the processes of accessioning and processing, which can minimize errors. These practical applications of 3D printing in the cytopathology laboratory demonstrate the utility of the design and printing process on improving aspects of the workflow in cytopathology laboratories to maximize efficiency, organization, and patient safety.
Novel and innovative ways to solve problems in the cytopathology laboratory are now possible with the increased access and decreased cost of three‐dimensional (3D) printers. In this report, an institution shares how expertise in 3D design and access to a 3D printing laboratory can help to create customized materials to address design issues in cytopathology specimen collection and processing.
To prospectively evaluate outcomes associated with use of a triple-lumen (TL) peripherally inserted central catheter (PICC) in the intensive care unit (ICU) setting.
Patients were prospectively ...enrolled in this HIPAA-compliant, institutional review board-approved study. Informed consent was obtained. All patients were in one hospital's ICUs and needed intermediate-term central venous access requiring three lumina. A 6-F tapered TL PICC was placed by a bedside nursing-based team with backup from the Interventional Radiology department. Placement complications, as well as long-term complications, were recorded. At catheter removal, ultrasonography (US) of the veins containing the TL PICC was performed to detect occult venous thrombosis. Regardless of indication for removal, catheters were sent for culture to detect colonization.
The study was stopped prematurely after 50 of a planned 167 patients were enrolled when a scheduled interim analysis detected a venous thrombosis rate that was considered unacceptably high by the study oversight committee (thrombosis was symptomatic in 20% of patients 10 of 50). Venous thrombosis (symptomatic or asymptomatic) was detected in 26 of 45 patients (58%; 95% confidence interval CI: 43%, 72%) examined with US. Documented catheter-related bloodstream infection did not occur (0%; 95% CI: 0%, 7%); colonization was detected in three of 29 catheter tips sent for culture (10%; 95% CI: 2%, 27%). Catheter malfunction and dislodgment occurred in one patient each.
The TL PICC design used in this study resulted in unacceptably high venous thrombosis rates. Even when used in a high-risk setting for infection (ie, the ICU), rates of clinically evident infection and colonization were absent and low, respectively.
This was a systematic review of existing literature.
The objective of this study was to evaluate the current state-of-the-art trends and utilization of machine learning in the field of spine surgery.
...The past decade has seen a rise in the clinical use of machine learning in many fields including diagnostic radiology and oncology. While studies have been performed that specifically pertain to spinal surgery, there have been relatively few aggregate reviews of the existing scientific literature as applied to clinical spine surgery.
This study utilized Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) methodology to review the scientific literature from 2009 to 2019 with syntax specific for machine learning and spine surgery applications. Specific data was extracted from the available literature including algorithm application, algorithms tested, database type and size, algorithm training method, and outcome of interest.
A total of 44 studies met inclusion criteria, of which the majority were level III evidence. Studies were grouped into 4 general types: diagnostic tools, clinical outcome prediction, surgical assessment tools, and decision support tools. Across studies, a wide swath of algorithms were used, which were trained across multiple disparate databases. There were no studies identified that assessed the ethical implementation or patient perceptions of machine learning in clinical care.
The results reveal the broad range of clinical applications and methods used to create machine learning algorithms for use in the field of spine surgery. Notable disparities exist in algorithm choice, database characteristics, and training methods. Ongoing research is needed to make machine learning operational on a large scale.
Machine learning promises to assist physicians with predictions of mortality and of other future clinical events by learning complex patterns from historical data, such as longitudinal electronic ...health records. Here we show that a convolutional neural network trained on raw pixel data in 812,278 echocardiographic videos from 34,362 individuals provides superior predictions of one-year all-cause mortality. The model's predictions outperformed the widely used pooled cohort equations, the Seattle Heart Failure score (measured in an independent dataset of 2,404 patients with heart failure who underwent 3,384 echocardiograms), and a machine learning model involving 58 human-derived variables from echocardiograms and 100 clinical variables derived from electronic health records. We also show that cardiologists assisted by the model substantially improved the sensitivity of their predictions of one-year all-cause mortality by 13% while maintaining prediction specificity. Large unstructured datasets may enable deep learning to improve a wide range of clinical prediction models.