Search NASA⌕ Search

SEARCH · Search NASA

Results for “Predictors”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 505 records · Page 28
All sourcesNASA NTRSDOE OSTI
DOE OSTI2011

Impact of the Radiation Boost on Outcomes After Breast-Conserving Surgery and Radiation

Purpose: We examined the impact of radiation tumor bed boost parameters in early-stage breast cancer on local control and cosmetic outcomes. Methods and Materials: A total of 3,186 women underwent postlumpectomy whole-breast radiation with a tumor bed boost for Tis to T2 breast cancer from 1970 to 2008. Boost parameters analyzed included size, energy, dose, and technique. Endpoints were local control, cosmesis, and fibrosis. The Kaplan-Meier method was used to estimate actuarial incidence, and a Cox proportional hazard model was used to determine independent predictors of outcomes on multivariate analysis (MVA). The median follow-up was 78 months (range, 1-305 months). Results: The crude cosmetic results were excellent in 54%, good in 41%, and fair/poor in 5% of patients. The 10-year estimate of an excellent cosmesis was 66%. On MVA, independent predictors for excellent cosmesis were use of electron boost, lower electron energy, adjuvant systemic therapy, and whole-breast IMRT. Fibrosis was reported in 8.4% of patients. The actuarial incidence of fibrosis was 11% at 5 years and 17% at 10 years. On MVA, independent predictors of fibrosis were larger cup size and higher boost energy. The 10-year actuarial local failure was 6.3%. There was no significant difference in local control by boost method, cut-out size, dose, or energy. Conclusions: Likelihood of excellent cosmesis or fibrosis are associated with boost technique, electron energy, and cup size. However, because of high local control and rare incidence of fair/poor cosmesis with a boost, the anatomy of the patient and tumor cavity should ultimately determine the necessary boost parameters.

62 RADIOLOGY AND NUCLEAR MEDICINE↗
DOE OSTI2012

Predictive Models for Pulmonary Function Changes After Radiotherapy for Breast Cancer and Lymphoma

Purpose: To propose multivariate predictive models for changes in pulmonary function tests ({Delta}PFTs) with respect to preradiotherapy (pre-RT) values in patients undergoing RT for breast cancer and lymphoma. Methods and Materials: A prospective study was designed to measure {Delta}PFTs of patients undergoing RT. Sixty-six patients were included. Spirometry, lung capacity (measured by helium dilution), and diffusing capacity of carbon monoxide tests were used to measure lung function. Two lung definitions were considered: paired lung vs. irradiated lung (IL). Correlation analysis of dosimetric parameters (mean lung dose and the percentage of lung volume receiving more than a threshold dose) and {Delta}PFTs was carried out to find the best dosimetric predictor. Chemotherapy, age, smoking, and the selected dose-volume parameter were considered as single and interaction terms in a multivariate analysis. Stability of results was checked by bootstrapping. Results: Both lung definitions proved to be similar. Modeling was carried out for IL. Acute and late damage showed the highest correlations with volumes irradiated above {approx}20 Gy (maximum R{sup 2} = 0.28) and {approx}40 Gy (maximum R{sup 2} = 0.21), respectively. RT alone induced a minor and transitory restrictive defect (p = 0.013). Doxorubicin-cyclophosphamide-paclitaxel (Taxol), when administered pre-RT, induced a late, large restrictive effect, independent of RT (p = 0.031). Bootstrap values confirmed the results. Conclusions: None of the dose-volume parameters was a perfect predictor of outcome. Thus, different predictor models for {Delta}PFTs were derived for the IL, which incorporated other nondosimetric parameters mainly through interaction terms. Late {Delta}PFTs seem to behave more serially than early ones. Large restrictive defects were demonstrated in patients pretreated with doxorubicin-cyclophosphamide-paclitaxel.

62 RADIOLOGY AND NUCLEAR MEDICINE↗
DOE OSTI2011

The Number of Pathologically Positive Lymph Nodes and Pathological Tumor Depth Predicts Prognosis in Patients With Poorly Differentiated Squamous Cell Carcinoma of the Oral Cavity

Purpose: The objective of this retrospective study was twofold: (1) to investigate prognostic factors for clinical outcomes in patients with poorly differentiated oral cavity squamous cell carcinoma and (2) to identify specific prognostic subgroups that may help to guide treatment decisions. Methods and Materials: We examined 102 patients with poorly differentiated oral cavity squamous cell carcinoma. All patients were followed for at least 24 months after surgery or until death. The 5-year rates of local control, neck control, distant metastasis, disease-free, disease-specific, and overall survival served as main outcome measures. Results: The 5-year rates were as follows: local control (79%), neck control (64%), distant metastases (27%), disease-free survival (48%), disease-specific survival (52%), and overall survival (42%). Multivariable analysis showed that the number of pathologically positive nodes ({>=}4 vs. {<=}3) was a significant predictor of neck control, distant metastasis, and disease-free, disease-specific, and overall survival rates. In addition, the presence of tumor depth of {>=}11 mm (vs. <11 mm) was a significant predictor of distant metastasis, disease-specific survival, and overall survival rates. The combination of the two predictors (26.5%, 27/102) was independently associated with poorer neck control (p = 0.0319), distant metastasis (p < 0.0001), and disease-free (p < 0.0001), disease-specific (p < 0.0001), and overall survival (p < 0.0001) rates. Conclusions: In patients with poorly differentiated oral cavity squamous cell carcinoma, the presence of at least 4 pathologically positive lymph nodes and of a pathological tumor depth {>=}11 mm identifies a subset of subjects with poor clinical outcomes. Patients carrying both risk factors are suitable candidates for the development of novel therapeutic approaches.

62 RADIOLOGY AND NUCLEAR MEDICINE↗
DOE OSTI2012

Long-Term Outcomes After High-Dose Postprostatectomy Salvage Radiation Treatment

Purpose: To review the impact of high-dose radiotherapy (RT) in the postprostatectomy salvage setting on long-term biochemical control and distant metastases-free survival, and to identify clinical and pathologic predictors of outcomes. Methods and Materials: During 1988-2007, 285 consecutive patients were treated with salvage RT (SRT) after radical prostatectomy. All patients were treated with either three-dimensional conformal RT or intensity-modulated RT. Two hundred seventy patients (95%) were treated to a dose {>=}66 Gy, of whom 205 (72%) received doses {>=}70 Gy. Eighty-seven patients (31%) received androgen-deprivation therapy as a component of their salvage treatment. All clinical and pathologic records were reviewed to identify treatment risk factors and response. Results: The median follow-up time after SRT was 60 months. Seven-year actuarial prostate-specific antigen (PSA) relapse-free survival and distant metastases-free survival were 37% and 77%, respectively. Independent predictors of biochemical recurrence were vascular invasion (p < 0.01), negative surgical margins (p < 0.01), presalvage PSA level >0.4 ng/mL (p < 0.01), androgen-deprivation therapy (p = 0.03), Gleason score {>=}7 (p = 0.02), and seminal vesicle involvement (p = 0.05). Salvage RT dose {>=}70 Gy was not associated with improvement in biochemical control. A doubling time <3 months was the only independent predictor of metastatic disease (p < 0.01). There was a trend suggesting benefit of SRT dose {>=}70 Gy in preventing clinical local failure in patients with radiographically visible local disease at time of SRT (7 years: 90% vs. 79.1%, p = 0.07). Conclusion: Salvage RT provides effective long-term biochemical control and freedom from metastasis in selected patients presenting with detectable PSA after prostatectomy. Androgen-deprivation therapy was associated with improvement in biochemical progression-free survival. Clinical local failures were rare but occurred most commonly in patients with greater burden of disease at time of SRT as reflected by either radiographic imaging or a greater PSA level. Salvage radiation doses {>=}70 Gy may ultimately be most beneficial in these patients, but this needs to be further studied.

62 RADIOLOGY AND NUCLEAR MEDICINE↗
DOE OSTI2013

Safety of {sup 90}Y Radioembolization in Patients Who Have Undergone Previous External Beam Radiation Therapy

Purpose: Previous external beam radiation therapy (EBRT) is theoretically contraindicated for yttrium-90 ({sup 90}Y) radioembolization (RE) because the liver has a lifetime tolerance to radiation before becoming vulnerable to radiation-induced liver disease. We analyzed the safety of RE as salvage treatment in patients who had previously undergone EBRT. Methods and Materials: Between June 2004 and December 2010, a total of 31 patients who had previously undergone EBRT were treated with RE. Three-dimensional treatment planning with dose–volume histogram (DVH) analysis of the liver was used to calculate the EBRT liver dose. Liver-related toxicities including RE-induced liver disease (REILD) were reviewed and classified according to Common Terminology Criteria for Adverse Events version 4.02. Results: The mean EBRT and RE liver doses were 4.40 Gy (range, 0-23.13 Gy) and 57.9 Gy (range, 27.0-125.9 Gy), respectively. Patients who experienced hepatotoxicity (≥grade2; n=12) had higher EBRT mean liver doses (7.96 ± 8.55 Gy vs 1.62 ± 3.39 Gy; P=.037), the only independent predictor in multivariate analysis. DVH analysis showed that the fraction of liver exposed to ≥30 Gy (V30) was the strongest predictor of hepatotoxicity (10.14% ± 12.75% vs 0.84% ± 3.24%; P=.006). All patients with V30 >13% experienced hepatotoxicity. Fatal REILD (n=2) occurred at the 2 highest EBRT mean liver doses (20.9 Gy and 23.1 Gy) but also at the highest cumulative liver doses (91.8 Gy and 149 Gy). Conclusions: Prior exposure of the liver to EBRT may lead to increased liver toxicity after RE treatment, depending on fractional liver exposure and dose level. The V30 was the strongest predictor of toxicity. RE appears to be safe for the treatment of hepatic malignancies only in patients who have had limited hepatic exposure to prior EBRT.

62 RADIOLOGY AND NUCLEAR MEDICINE↗
DOE OSTI2013

MRI Predictive Factors for Tumor Response in Rectal Cancer Following Neoadjuvant Chemoradiation Therapy - Implications for Induction Chemotherapy?

Purpose: Clinical and magnetic resonance imaging (MRI) characteristics at baseline and following chemoradiation therapy (CRT) most strongly associated with histopathologic response were investigated and survival outcomes evaluated in accordance with imaging and pathological response. Methods and Materials: Responders were defined as mrT3c/d-4 downstaged to ypT0-2 on pathology or low at risk mrT2 downstaged to ypT1 or T0. Multivariate logistic regression of baseline and posttreatment MRI: T, N, extramural venous invasion (EMVI), circumferential resection margin, craniocaudal length <5 cm, and MRI tumor height ≤5 cm were used to identify independent predictor(s) for response. An association between induction chemotherapy and EMVI status was analyzed. Survival outcomes for pathologic and MRI responders and nonresponders were analyzed. Results: Two hundred eighty-one patients were eligible; 114 (41%) patients were pathology responders. Baseline MRI negative EMVI (odds ratio 2.94, P=.007), tumor height ≤5 cm (OR 1.96, P=.02), and mrEMVI status change (positive to negative) following CRT (OR 3.09, P<.001) were the only predictors for response. There was a strong association detected between induction chemotherapy and ymrEMVI status change after CRT (OR 9.0, P<.003). ymrT0-2 gave a positive predictive value of 80% and OR of 9.1 for ypT0-2. ymrN stage accuracy of ypN stage was 75%. Three-year disease-free survival for pathology and MRI responders were similar at 80% and 79% and significantly better than poor responders. Conclusions: Tumor height and mrEMVI status are more important than baseline size and stage of the tumor as predictors of response to CRT. Both MRI- and pathologic-defined responders have significantly improved survival. “Good response” to CRT in locally advanced rectal cancer with ypT0-2 carries significantly better 3-year overall survival and disease-free survival. Use of induction chemotherapy for improving mrEMVI status and knowledge of MRI predictive factors could be taken into account in the pursuit of individualized neoadjuvant treatments for patients with rectal cancer.

62 RADIOLOGY AND NUCLEAR MEDICINE↗
DOE OSTI2015

Dosimetric Inhomogeneity Predicts for Long-Term Breast Pain After Breast-Conserving Therapy

Purpose: The objective of this cross-sectional study was to characterize long-term breast pain in patients undergoing breast-conserving surgery and radiation (BCT) and to identify predictors of this pain. Methods and Materials: We identified 355 eligible patients with Tis-T2N0M0 breast cancer who underwent BCT in 2007 to 2011, without recurrent disease. A questionnaire derived from the Late Effects Normal Tissue Task Force (LENT) Subjective, Objective, Management, Analytic (SOMA) scale was mailed with 7 items detailing the severity, frequency, duration, and impact of ipsilateral breast pain over the previous 2 weeks. A logistic regression model identified predictors of long-term breast pain based on questionnaire responses and patient, disease, and treatment characteristics. Results: The questionnaire response rate was 80% (n=285). One hundred thirty-five patients (47%) reported pain in the treated breast, with 19 (14%) having pain constantly or at least daily; 15 (11%) had intense pain. The pain interfered with daily activities in 11 patients (8%). Six patients (4%) took analgesics for breast pain. Fourteen (10%) thought that the pain affected their quality of life. On univariable analysis, volume of breast tissue treated to ≥105% of the prescribed dose (odds ratio [OR] 1.001 per cc, 95% confidence interval [CI] 1.000-1.002; P=.045), volume treated to ≥110% (OR 1.009 per cc, 95% CI 1.002-1.016; P=.012), hormone therapy use (OR 1.95, 95% CI 1.12-3.39; P=.02), and other sites of pain (OR 1.79, 95% CI 1.05-3.07; P=.03) predicted for long-term breast pain. On multivariable analysis, volume ≥110% (OR 1.01 per cc, 95% CI 1.003-1.017; P=.007), shorter time since treatment (OR 0.98 per month, 95% CI 0.96-0.998; P=.03), and hormone therapy (OR 1.84, 95% CI 1.05-3.25; P=.03) were independent predictors of pain. Conclusion: Long-term breast pain was common after BCT. Although nearly half of patients had pain, most considered it tolerable. Dosimetric inhomogeneity independently predicted for pain and should be minimized to the greatest extent possible.

62 RADIOLOGY AND NUCLEAR MEDICINE↗
DOE OSTI2014

Racial Variations in Radiation-Induced Skin Toxicity Severity: Data From a Prospective Cohort Receiving Postmastectomy Radiation

Purpose: Radiation-induced skin toxicity is one of the most symptomatic side effects of postmastectomy radiation therapy (PMRT). We sought to determine whether the severity of acute skin toxicity was greater in black patients in a prospective cohort receiving PMRT and to identify other predictors of more severe skin toxicity. Methods and Materials: We evaluated the first 110 patients in an ongoing prospective study assessing radiation-induced skin toxicity in patients receiving PMRT. We recorded patient demographics, body mass index (BMI), and disease and treatment characteristics. Logistic regression analyses were conducted to evaluate the effect of potential predictors on the risk of skin toxicity. Results: A total of 23.6% respondents self-identified as black, 5.5% as non-Hispanic white, 69.1% as Hispanic white, and 1.8% as other; 57% were postmenopausal, and 70.9% had BMI of >25. Median chest wall dose was 50 Gy, and mastectomy scar dose was 60 Gy. Most patients, 95.5%, were treated with a 0.5-cm bolus throughout treatment. There were no significant differences in patient characteristics in black versus non-black patients. At RT completion, moist desquamation was more common in black patients (73.1% vs 47.6%, respectively, P=.023), in postmenopausal patients (63.5% vs 40.4%, respectively, P=.016), and in those with BMI of ≥25 (60.3% vs 37.5%, respectively, P=.030). On multivariate analysis, the effects of black race (odds ratio [OR] = 7.46, P=.031), BMI ≥25 (OR = 2.95, P=.043) and postmenopausal status (OR = 8.26, P=.004) remained significant risk factors for moist desquamation. Conclusions: In this prospectively followed, racially diverse cohort of breast cancer patients receiving PMRT delivered in a uniform fashion, including the routine use of chest wall boost and bolus, black race, higher BMI, and postmenopausal status emerged as significant predictors of moist desquamation. There was a high frequency of moist desquamation, particularly in those patients with elevated risk. Continued study of patient selection for chest wall boost and bolus as well improved skin toxicity management strategies are needed.

62 RADIOLOGY AND NUCLEAR MEDICINE↗
DOE OSTI2015

The Prognostic Value of Plasma Epstein-Barr Viral DNA and Tumor Response to Neoadjuvant Chemotherapy in Advanced-Stage Nasopharyngeal Carcinoma

Purpose: To explore the prognostic value of the plasma load of Epstein-Barr viral (EBV) DNA and the tumor response to neoadjuvant chemotherapy (NACT) in advanced-stage nasopharyngeal carcinoma (NPC). Patients and Methods: In all, 185 consecutive patients with stage III to IVb NPC treated with NACT followed by concurrent chemoradiation therapy (CCRT) were prospectively enrolled. The primary endpoint was progression-free survival (PFS), and the secondary endpoints included locoregional relapse–free survival (LRFS) and distant metastasis–free survival (DMFS). Results: EBV DNA was detected in 165 (89%) patients before treatment but was undetectable in 127 (69%) patients after NACT. Detectable EBV DNA levels after NACT were correlated with poor prognosis (3-year PFS 71.8% vs 85.2%, P=.008 and 3-year DMFS 82.5% vs 92.3%, P=.013). An unsatisfactory tumor response (stable disease or disease progression) after NACT was also correlated with poor clinical outcome (3-year PFS 71.1% vs 85.9%, P=.005 and 3-year LRFS 82.7% vs 93.5%, P=.012). Multivariate analysis showed that the EBV DNA level after NACT (hazard ratio [HR] 2.31, 95% CI 1.18-4.54, P=.015) and the tumor response to NACT (HR 2.84, 95% CI 1.42-5.67, P=.003) were both significant prognostic factors for PFS. Multivariate analysis also showed that EBV DNA after NACT was the only significant predictor of DMFS (HR 2.99, 95% CI 1.25-7.15, P=.014) and that tumor response to NACT was the only significant predictor of LRFS (HR 3.31, 95% CI 1.21-9.07, P=.020). Conclusion: Detectable EBV DNA levels and an unsatisfactory tumor response (stable disease or disease progression) after NACT serve as predictors of poor prognosis for patients with advanced-stage NPC. These findings will facilitate further risk stratification, early treatment modification, or both before CCRT.

62 RADIOLOGY AND NUCLEAR MEDICINE↗
DOE OSTI2016

Clinical Indicators of Psychosocial Distress Predict for Acute Radiation-Induced Fatigue in Patients Receiving Adjuvant Radiation Therapy for Breast Cancer: An Analysis of Patient-Reported Outcomes

Purpose: To assess the magnitude and predictors of patient-reported fatigue among breast cancer patients receiving radiation therapy (RT). Methods and Materials: Patients receiving breast RT completed a survey querying fatigue at each weekly on-treatment visit. Patient-reported fatigue severity and interference was assessed on an ordinal scale of 0 to 4, using a validated scoring system. Baseline anxiety and depression scores were also obtained. The kinetics of mean fatigue scores per week and the maximum fatigue scores over the course of the entire treatment were assessed, and clinical predictors were identified by univariate and multivariate regression. Results: The average fatigue severity and interference scores were 0.6 and 0.46. The average fatigue scores increased to an equivalent extent from week to week, with expected increases of 0.99 in fatigue severity and 0.85 in interference over 7 weeks. Patients treated with hypofractionated RT (HF-RT) versus conventionally fractionated RT (CF-RT) had significantly fewer maximum fatigue severity or interference scores that were >2 (ie, severe or very severe; 29% vs 10% for severity, and 26% vs 8% for interference, P<.01). Age ≤45 years, presence of psychiatric/pain-related comorbidities, and baseline sadness and anxiety severity were predictive of average and maximum fatigue scores (P<.05), but variables related to treatment intensity (eg, mastectomy vs lumpectomy, chemotherapy use, radiation target volumes) and other host factors (working, children, marital status, proximity to RT facility) were not. Conclusion: Patient-reported fatigue modestly increases over RT courses, with less maximum fatigue reported with HF-RT. Younger age and baseline sadness, anxiety, and psychiatric/pain-related comorbidities are powerful predictors of fatigue, whereas other factors, such as treatment intensity, are not. Future studies will investigate interventions for patients at high risk for fatigue.

62 RADIOLOGY AND NUCLEAR MEDICINE↗
DOE OSTI2016

Pitfalls in Prediction Modeling for Normal Tissue Toxicity in Radiation Therapy: An Illustration With the Individual Radiation Sensitivity and Mammary Carcinoma Risk Factor Investigation Cohorts

Purpose: To identify the main causes underlying the failure of prediction models for radiation therapy toxicity to replicate. Methods and Materials: Data were used from two German cohorts, Individual Radiation Sensitivity (ISE) (n=418) and Mammary Carcinoma Risk Factor Investigation (MARIE) (n=409), of breast cancer patients with similar characteristics and radiation therapy treatments. The toxicity endpoint chosen was telangiectasia. The LASSO (least absolute shrinkage and selection operator) logistic regression method was used to build a predictive model for a dichotomized endpoint (Radiation Therapy Oncology Group/European Organization for the Research and Treatment of Cancer score 0, 1, or ≥2). Internal areas under the receiver operating characteristic curve (inAUCs) were calculated by a naïve approach whereby the training data (ISE) were also used for calculating the AUC. Cross-validation was also applied to calculate the AUC within the same cohort, a second type of inAUC. Internal AUCs from cross-validation were calculated within ISE and MARIE separately. Models trained on one dataset (ISE) were applied to a test dataset (MARIE) and AUCs calculated (exAUCs). Results: Internal AUCs from the naïve approach were generally larger than inAUCs from cross-validation owing to overfitting the training data. Internal AUCs from cross-validation were also generally larger than the exAUCs, reflecting heterogeneity in the predictors between cohorts. The best models with largest inAUCs from cross-validation within both cohorts had a number of common predictors: hypertension, normalized total boost, and presence of estrogen receptors. Surprisingly, the effect (coefficient in the prediction model) of hypertension on telangiectasia incidence was positive in ISE and negative in MARIE. Other predictors were also not common between the 2 cohorts, illustrating that overcoming overfitting does not solve the problem of replication failure of prediction models completely. Conclusions: Overfitting and cohort heterogeneity are the 2 main causes of replication failure of prediction models across cohorts. Cross-validation and similar techniques (eg, bootstrapping) cope with overfitting, but the development of validated predictive models for radiation therapy toxicity requires strategies that deal with cohort heterogeneity.

62 RADIOLOGY AND NUCLEAR MEDICINE↗
DOE OSTI2017

Postmastectomy Radiation Therapy Is Associated With Improved Survival in Node-Positive Male Breast Cancer: A Population Analysis

Purpose: Because of its rarity, there are no randomized trials investigating postmastectomy radiation therapy (PMRT) in male breast cancer. This study retrospectively examines the impact of PMRT in male breast cancer patients in the National Cancer Institute's Surveillance, Epidemiology, and End Results (SEER) database. Methods and Materials: The SEER database 8.3.2 was queried for men ages 20+ with a diagnosis of localized or regional nonmetastatic invasive ductal/lobular carcinoma from 1998 to 2013. Included patients were treated by modified radical mastectomy (MRM), with or without adjuvant external beam radiation. Univariate and multivariate analyses evaluated predictors for PMRT use after MRM. Kaplan-Meier overall survival (OS) curves of the entire cohort and a case-matched cohort were calculated and compared by the log-rank test. Cox regression was used for multivariate survival analyses. Results: A total of 1933 patients were included in the unmatched cohort. There was no difference in 5-year OS between those who received PMRT and those who did not (78% vs 77%, respectively, P=.371); however, in the case-matched analysis, PMRT was associated with improved OS at 5 years (83% vs 54%, P<.001). On subset analysis of the unmatched cohort, PMRT was associated with improved OS in men with 1 to 3 positive nodes (5-year OS 79% vs 72% P=.05) and those with 4+ positive nodes (5-year OS 73% vs 53% P<.001). On multivariate analysis of the unmatched cohort, independent predictors for improved OS were use of PMRT: HR=0.551 (0.412-0.737) and estrogen receptor–positive disease: HR=0.577 (0.339-0.983). Predictors for a survival detriment were higher grade 3/4: HR=1.825 (1.105-3.015), larger tumor T2: HR=1.783 (1.357-2.342), T3/T4: HR=2.683 (1.809-3.978), higher N-stage: N1 HR=1.574 (1.184-2.091), N2/N3: HR=2.328 (1.684-3.218), black race: HR=1.689 (1.222-2.336), and older age 81+: HR=4.164 (1.497-11.582). Conclusions: There may be a survival benefit with the addition of PMRT for male breast cancer with node-positive disease.

62 RADIOLOGY AND NUCLEAR MEDICINE↗
DOE OSTI2023

Uncertainty quantification of a deep learning fuel property prediction model

Deep learning models are being widely used in the field of combustion. Given the black-box nature of typical neural network based models, uncertainty quantification (UQ) is critical to ensure the reliability of predictions as well as the training datasets, and for a principled quantification of noise and its various sources. Deep learning surrogate models for predicting properties of chemical compounds and mixtures have been recently shown to be promising for enabling data-driven fuel design and optimization, with the ultimate goal of improving efficiency and lowering emissions from combustion engines. In this study, UQ is performed for a multi-task deep learning model that simultaneously predicts the research octane number (RON), Motor Octane Number (MON), and Yield Sooting Index (YSI) of pure components and multicomponent blends. The deep learning model is comprised of three smaller networks: Extractor 1, Extractor 2, and Predictor, and a mixing operator. The molecular fingerprints of individual components are encoded via Extractor 1 and Extractor 2, the mixing operator generates fingerprints for mixtures/blends based on linear mixing operation, and the predictor maps the fingerprint to the target properties. Two different classes of UQ methods, Monte Carlo ensemble methods and Bayesian neural networks (BNNs), are employed for quantifying the epistemic uncertainty. Combinations of Bernoulli and Gaussian distributions with DropConnect and DropOut techniques are explored as ensemble methods. All the DropConnect, DropOut and Bayesian layers are applied to the predictor network. Aleatoric uncertainty is modeled by assuming that each data point has an independent uncertainty associated with it. The results of the UQ study are further analyzed to compare the performance of BNN and ensemble methods. Although this study is confined to UQ of fuel property prediction, the methodologies are applicable to other deep learning frameworks that are being widely used in the combustion community.

33 ADVANCED PROPULSION SYSTEMS↗
DOE OSTI2022

Aboveground biomass density models for NASA’s Global Ecosystem Dynamics Investigation (GEDI) lidar mission

NASA's Global Ecosystem Dynamics Investigation (GEDI) is collecting spaceborne full waveform lidar data with a primary science goal of producing accurate estimates of forest aboveground biomass density (AGBD). This paper presents the development of the models used to create GEDI's footprint-level (~25 m) AGBD (GEDI04_A) product, including a description of the datasets used and the procedure for final model selection. The data used to fit our models are from a compilation of globally distributed spatially and temporally coincident field and airborne lidar datasets, whereby we simulated GEDI-like waveforms from airborne lidar to build a calibration database. We used this database to expand the geographic extent of past waveform lidar studies, and divided the globe into four broad strata by Plant Functional Type (PFT) and six geographic regions. GEDI's waveform-to-biomass models take the form of parametric Ordinary Least Squares (OLS) models with simulated Relative Height (RH) metrics as predictor variables. From an exhaustive set of candidate models, we selected the best input predictor variables, and data transformations for each geographic stratum in the GEDI domain to produce a set of comprehensive predictive footprint-level models. We found that model selection frequently favored combinations of RH metrics at the 98th, 90th, 50th, and 10th height above ground-level percentiles (RH98, RH90, RH50, and RH10, respectively), but that inclusion of lower RH metrics (e.g. RH10) did not markedly improve model performance. Second, forced inclusion of RH98 in all models was important and did not degrade model performance, and the best performing models were parsimonious, typically having only 1-3 predictors. Third, stratification by geographic domain (PFT, geographic region) improved model performance in comparison to global models without stratification. Fourth, for the vast majority of strata, the best performing models were fit using square root transformation of field AGBD and/or height metrics. There was considerable variability in model performance across geographic strata, and areas with sparse training data and/or high AGBD values had the poorest performance. These models are used to produce global predictions of AGBD, but will be improved in the future as more and better training data become available.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2015

Public attitudes regarding large-scale solar energy development in the U.S.

Using data collected from both a National sample as well as an oversample in U.S. Southwest, we examine public attitudes toward the construction of utility-scale solar facilities in the U.S. as well as development in one’s own county. Our multivariate analyses assess demographic and sociopsychological factors as well as context in terms of proximity of proposed project by considering the effect of predictors for respondents living in the Southwest versus those from a National sample.We find that the predictors, and impact of the predictors, related to support and opposition to solar development vary in terms of psychological and physical distance. Overall, for respondents living in the U.S. Southwest we find that environmentalism, belief that developers receive too many incentives, and trust in project developers to be significantly related to support and opposition to solar development, in general. When Southwest respondents consider large-scale solar development in their county, the influence of these variables changes so that property value, race, and age only yield influence. Differential effects occur for respondents of our National sample.We believe our findings to be relevant for those outside the U.S. due to the considerable growth PV solar has experienced in the last decade, especially in China, Japan, Germany, and the U.S.

99 GENERAL AND MISCELLANEOUS↗
DOE OSTI2021

CyProduct: A software tool for accurately predicting the byproducts of human cytochrome P450 metabolism

In silico metabolism prediction is a cheminformatic task of autonomously predicting the set of metabolic byproducts produced from a specified molecule and a set of enzymes or reactions. Here we describe a novel machine-learned in silico cytochrome P450 (CYP450) metabolism prediction suite, called CyProduct, that accurately predicts metabolic byproducts for a specified molecule and a human CYP450 isoform. It includes three modules: (1) CypReact, a tool that predicts if the query compound reacts with a given CYP450 enzyme; (2) CypBoM, a tool that accurately predicts the “bond site” of the reaction (i.e., which specific bonds within the query molecule react with the CYP isoform); and (3) MetaboGen, a tool that generates the metabolic byproducts based on CypBoM’s bond-site prediction. CyProduct predicts metabolic biotransformation products for each of the nine most important human CYP450 enzymes. CypBoM uses an important new concept called “Bond of Metabolism” (BoM), which extends the traditional “Site of Metabolism" (SoM) by specifying the information about the set of chemical bonds that is modified or formed in a metabolic reaction (rather than the specific atom). We created a BoM database for 3487 CYP450-mediated Phase I reactions, then used this to train the CypBoM Predictor to predict the reactive bond locations on substrate molecules. CypBoM Predictor’s cross-validated Jaccard score for reactive bond prediction ranged from 0.380 to 0.452 over the nine CYP450 enzymes. Over variants of a test set of 72 known CYP450 substrates and 30 non-reactants, CyProduct outperformed the other packages -- including ADMET Predictor, BioTransformer and GLORY -- by an average of 200% (wrt Jaccard score) in terms of predicting metabolites. The CyProduct suite and the datasets are freely available at https://bitbucket.org/wishartlab/cyproduct/src/master/.

Machine learning, Cytochrome P450, Metabolism pred↗
DOE OSTI2018

Prognostic Power of Extreme Rainfall Scaling Formulas Across Space and Time Scales

Some studies documenting changes in extreme precipitation use scaling formulas to approximate the large percentiles of the rainfall distribution from average dynamical and thermodynamical variables, called predictors. Here we instead assess the performance of these formulas as approximations to the rain rates in individual events. We evaluate the accuracies of the scaling relationships as functions of spatial and temporal scales by analyzing tropical rainfall in a superparameterized model. Relationships using full vertical profiles of the predictors are more accurate than those using their values at specific vertical levels because they better characterize the specific dynamics of each event. Both types of scaling relationships perform well over a range of length scales from 200 to 2,000 km and time scales from an hour to a week, and their precision is higher in the case of simulations with superparameterization than with parameterized convection. Uncertainties emerging from the local use of the scaling relationships suggest that they may only characterize the intensification of individual extremes for a warming of 4–5 K or larger. Finally, we argue that these formulas can be used to reconstruct the tail of the rainfall distribution directly from its predictors without prior information on P. While scalings have been used as diagnostic equations conditioned on the occurrence of extreme rainfall, they are actually able to mimic the prognostic behavior of climate model parameterizations on a variety of scales when estimating the intensity, frequency, and spatial patterns of extremes.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2022

Short–Term Forecasting of Wind Gusts at Airports Across CONUS Using Machine Learning

Short–term forecasting of wind gusts, particularly those of higher intensity, is of great societal importance but is challenging due to the presence of multiple gust generation mechanisms. Wind gust observations from eight high–passenger–volume airports across the continental United States (CONUS) are summarized and used to develop predictive models of wind gust occurrence and magnitude. These short–term (same hour) forecast models are built using multiple logistic and linear regression, as well as artificial neural networks (ANNs) of varying complexity. A suite of 19 upper–air predictors drawn from the ERA5 reanalysis and an autoregressive (AR) term are used. Stepwise procedures instruct predictor selection, and resampling is used to quantify model stability. All models are developed separately for the warm (April–September) and cold (October–March) seasons. Results show that ANNs of 3–5 hidden layers (HLs) generally exhibit higher hit rates than logistic regression models and also improve skill with respect to wind gust magnitudes. However, deeper networks with more HLs increase false alarm rates in occurrence models and mean absolute error in magnitude models due to model overfitting. For model skill, inclusion of the AR term is critical while the majority of the remaining skill derives from wind speeds and lapse rates. A predictive ceiling is also clearly demonstrated, particularly for the strong and damaging gust magnitudes, which appears to be partially due to ERA5 predictor characteristics and the presence of mixed wind climates.

54 ENVIRONMENTAL SCIENCES↗