| 1. |
- Sherratt, K., et al.
(författare)
-
Predictive performance of multi-model ensemble forecasts of COVID-19 across European nations
- 2023
-
Ingår i: eLIFE. - : eLife Sciences Publications Ltd. - 2050-084X. ; 12
-
Tidskriftsartikel (refereegranskat)abstract
- Methods: We used open-source tools to develop a public European COVID-19 Forecast Hub. We invited groups globally to contribute weekly forecasts for COVID-19 cases and deaths reported by a standardised source for 32 countries over the next 1–4 weeks. Teams submitted forecasts from March 2021 using standardised quantiles of the predictive distribution. Each week we created an ensemble forecast, where each predictive quantile was calculated as the equally-weighted average (initially the mean and then from 26th July the median) of all individual models’ predictive quantiles. We measured the performance of each model using the relative Weighted Interval Score (WIS), comparing models’ forecast accuracy relative to all other models. We retrospectively explored alternative methods for ensemble forecasts, including weighted averages based on models’ past predictive performance.Results: Over 52 weeks, we collected forecasts from 48 unique models. We evaluated 29 models’ forecast scores in comparison to the ensemble model. We found a weekly ensemble had a consistently strong performance across countries over time. Across all horizons and locations, the ensemble performed better on relative WIS than 83% of participating models’ forecasts of incident cases (with a total N=886 predictions from 23 unique models), and 91% of participating models’ forecasts of deaths (N=763 predictions from 20 models). Across a 1–4 week time horizon, ensemble performance declined with longer forecast periods when forecasting cases, but remained stable over 4 weeks for incident death forecasts. In every forecast across 32 countries, the ensemble outperformed most contributing models when forecasting either cases or deaths, frequently outperforming all of its individual component models. Among several choices of ensemble methods we found that the most influential and best choice was to use a median average of models instead of using the mean, regardless of methods of weighting component forecast models.Conclusions: Our results support the use of combining forecasts from individual models into an ensemble in order to improve predictive performance across epidemiological targets and populations during infectious disease epidemics. Our findings further suggest that median ensemble methods yield better predictive performance more than ones based on means. Our findings also highlight that forecast consumers should place more weight on incident death forecasts than incident case forecasts at forecast horizons greater than 2 weeks.
|
|
| 2. |
- Eroglu, Z., et al.
(författare)
-
Outcomes with adjuvant anti-PD-1 therapy in patients with sentinel lymph node-positive melanoma without completion lymph node dissection
- 2022
-
Ingår i: Journal for Immunotherapy of Cancer. - : BMJ. - 2051-1426. ; 10:8
-
Tidskriftsartikel (refereegranskat)abstract
- Until recently, most patients with sentinel lymph node-positive (SLN+) melanoma underwent a completion lymph node dissection (CLND), as mandated in published trials of adjuvant systemic therapies. Following multicenter selective lymphadenectomy trial-II, most patients with SLN+ melanoma no longer undergo a CLND prior to adjuvant systemic therapy. A retrospective analysis of clinical outcomes in SLN+ melanoma patients treated with adjuvant systemic therapy after July 2017 was performed in 21 international cancer centers. Of 462 patients who received systemic adjuvant therapy, 326 patients received adjuvant anti-PD-1 without prior immediate (IM) CLND, while 60 underwent IM CLND. With median follow-up of 21 months, 24-month relapse-free survival (RFS) was 67% (95% CI 62% to 73%) in the 326 patients. When the patient subgroups who would have been eligible for the two adjuvant anti-PD-1 clinical trials mandating IM CLND were analyzed separately, 24-month RFS rates were 64%, very similar to the RFS rates from those studies. Of these no-CLND patients, those with SLN tumor deposit >1 mm, stage IIIC/D and ulcerated primary had worse RFS. Of the patients who relapsed on adjuvant anti-PD-1, those without IM CLND had a higher rate of relapse in the regional nodal basin than those with IM CLND (46% vs 11%). Therefore, 55% of patients who relapsed without prior CLND underwent surgery including therapeutic lymph node dissection (TLND), with 30% relapsing a second time; there was no difference in subsequent relapse between patients who received observation vs secondary adjuvant therapy. Despite the increased frequency of nodal relapses, adjuvant anti-PD-1 therapy may be as effective in SLN+ pts who forego IM CLND and salvage surgery with TLND at relapse may be a viable option for these patients.
|
|
| 3. |
|
|
| 4. |
- Fairweather-Tait, S, et al.
(författare)
-
The usefulness of in vitro models to predict the bioavailability of iron and zinc: A consensus statement from the HarvestPlus Expert Consultation
- 2005
-
Ingår i: International Journal for Vitamin and Nutrition Research. - : Hogrefe Publishing Group. - 0300-9831 .- 1664-2821. ; 75:6, s. 371-374
-
Tidskriftsartikel (refereegranskat)abstract
- A combination of dietary and host-related factors determines iron and zinc absorption, and several in vitro methods have been developed as preliminary screening tools for assessing bioavailability. An expert committee has reviewed evidence for their usefulness and reached a consensus. Dialyzability (with and without simulated digestion) gives some useful information but cannot predict the correct magnitude of response and may sometimes predict the wrong direction of response. Caco-2 cell systems (with and without simulated digestion) have been developed for iron availability, but the magnitude of different effects does not always agree with results obtained in human volunteers, and the data for zinc are too limited to draw conclusions about the validity of the method. Caco-2 methodologies vary significantly between laboratories and require experienced technicians and good quality cell culture facilities to obtain reproducible results. Algorithms can provide semi-quantitative information enabling diets to be classified as high, moderate, or low bioavailability. While in vitro methods can be used to generate ideas and develop hypotheses, they cannot be used alone for important decisions concerning; food fortification policy, selection of varieties for plant breeding programs, or for new product development in the food industry. Ultimately human studies are required for such determinations.
|
|
| 5. |
|
|
| 6. |
- Lukasczyk, J., et al.
(författare)
-
Report of the TopoInVis TTK Hackathon : Experiences, Lessons Learned, and Perspectives
- 2021
-
Ingår i: Mathematics and Visualization. - Cham : Springer Nature. ; , s. 359-373, s. 359-373
-
Konferensbidrag (refereegranskat)abstract
- This paper documents the organization, the execution, and the results of the Topology ToolKit (TTK) hackathon that took place at the TopoInVis 2019 conference. The primary goal of the hackathon was to promote TTK in our research community as a unified software development platform for topology-based data analysis algorithms. To this end, participants were first introduced to the structure and capabilities of TTK, and then worked on their own TTK-related projects while being mentored by senior TTK developers. Notable outcomes of the hackathon were first steps towards Python and Docker packages, further integration of TTK in Inviwo, the extension of TTK with new algorithms, and the discovery of current limitations of TTK as well as future development directions.
|
|
| 7. |
- Sreevalsan-Nair, Jaya, et al.
(författare)
-
Human-Guided Enhancement of a Stochastic Local Search: Visualization and Adjustment of 3D Pheromone
- 2007
-
Konferensbidrag (refereegranskat)abstract
- In this paper, we describe user interaction with an optimization algorithm via a sophisticated visualization interface that we created for this purpose. Our primary interest is the tool itself. We demonstrate that a user wielding this tool can find ways to improve the performance of an ant colony optimization (ACO) algorithm as applied to a problem of finding 3D paths in the presence of impediments [14]. One part of a solution method can be to find a path on a grid. Of course, there are near linear time algorithms for the shortest path that have been applied to problems that are quite large. However, for a grid in three dimensions with arcs on the axes and diagonals, the problems can become extremely large as resolution is increased and heuristics thus make sense (see, e.g., [6] for state-of-the art algorithms where pre-processing is possible). Ant colony optimization (see, e.g., [4,5]) is ideally suited to such a problem.
|
|
| 8. |
- Yuh, Esther L, et al.
(författare)
-
Pathological computed tomography features associated with adverse outcomes after mild traumatic brain injury : A TRACK-TBI study with external validation in CENTER-TBI.
- 2021
-
Ingår i: JAMA Neurology. - : American Medical Association (AMA). - 2168-6149 .- 2168-6157. ; 78:9, s. 1137-1148
-
Tidskriftsartikel (refereegranskat)abstract
- IMPORTANCE: A head computed tomography (CT) with positive results for acute intracranial hemorrhage is the gold-standard diagnostic biomarker for acute traumatic brain injury (TBI). In moderate to severe TBI (Glasgow Coma Scale [GCS] scores 3-12), some CT features have been shown to be associated with outcomes. In mild TBI (mTBI; GCS scores 13-15), distribution and co-occurrence of pathological CT features and their prognostic importance are not well understood.OBJECTIVE: To identify pathological CT features associated with adverse outcomes after mTBI.DESIGN, SETTING, AND PARTICIPANTS: The longitudinal, observational Transforming Research and Clinical Knowledge in Traumatic Brain Injury (TRACK-TBI) study enrolled patients with TBI, including those 17 years and older with GCS scores of 13 to 15 who presented to emergency departments at 18 US level 1 trauma centers between February 26, 2014, and August 8, 2018, and underwent head CT imaging within 24 hours of TBI. Evaluations of CT imaging used TBI Common Data Elements. Glasgow Outcome Scale-Extended (GOSE) scores were assessed at 2 weeks and 3, 6, and 12 months postinjury. External validation of results was performed via the Collaborative European NeuroTrauma Effectiveness Research in Traumatic Brain Injury (CENTER-TBI) study. Data analyses were completed from February 2020 to February 2021.EXPOSURES: Acute nonpenetrating head trauma.MAIN OUTCOMES AND MEASURES: Frequency, co-occurrence, and clustering of CT features; incomplete recovery (GOSE scores <8 vs 8); and an unfavorable outcome (GOSE scores <5 vs ≥5) at 2 weeks and 3, 6, and 12 months.RESULTS: In 1935 patients with mTBI (mean [SD] age, 41.5 [17.6] years; 1286 men [66.5%]) in the TRACK-TBI cohort and 2594 patients with mTBI (mean [SD] age, 51.8 [20.3] years; 1658 men [63.9%]) in an external validation cohort, hierarchical cluster analysis identified 3 major clusters of CT features: contusion, subarachnoid hemorrhage, and/or subdural hematoma; intraventricular and/or petechial hemorrhage; and epidural hematoma. Contusion, subarachnoid hemorrhage, and/or subdural hematoma features were associated with incomplete recovery (odds ratios [ORs] for GOSE scores <8 at 1 year: TRACK-TBI, 1.80 [95% CI, 1.39-2.33]; CENTER-TBI, 2.73 [95% CI, 2.18-3.41]) and greater degrees of unfavorable outcomes (ORs for GOSE scores <5 at 1 year: TRACK-TBI, 3.23 [95% CI, 1.59-6.58]; CENTER-TBI, 1.68 [95% CI, 1.13-2.49]) out to 12 months after injury, but epidural hematoma was not. Intraventricular and/or petechial hemorrhage was associated with greater degrees of unfavorable outcomes up to 12 months after injury (eg, OR for GOSE scores <5 at 1 year in TRACK-TBI: 3.47 [95% CI, 1.66-7.26]). Some CT features were more strongly associated with outcomes than previously validated variables (eg, ORs for GOSE scores <5 at 1 year in TRACK-TBI: neuropsychiatric history, 1.43 [95% CI .98-2.10] vs contusion, subarachnoid hemorrhage, and/or subdural hematoma, 3.23 [95% CI 1.59-6.58]). Findings were externally validated in 2594 patients with mTBI enrolled in the CENTER-TBI study.CONCLUSIONS AND RELEVANCE: In this study, pathological CT features carried different prognostic implications after mTBI to 1 year postinjury. Some patterns of injury were associated with worse outcomes than others. These results support that patients with mTBI and these CT features need TBI-specific education and systematic follow-up.
|
|
