skip to main content
OSTI.GOV title logo U.S. Department of Energy
Office of Scientific and Technical Information

Title: A Toxicity Evaluation and Predictive System Based on Neural Networks and Wavelets

Abstract

A computational approach has been developed for performing efficient and reasonably accurate toxicity evaluation and prediction. The approach is based on computational neural networks linked to modern computational chemistry and wavelet methods. In this paper we present details of this approach and results demonstrating its accuracy and flexibility for predicting diverse biological endpoints including metabolic processes, mode of action, and hepato- and neurotoxicity. The approach also can be used for automatic processing of microarray data to predict modes of action.

Authors:
 [1];  [1];  [2];  [2];  [1];  [2];  [2];  [1];  [2]
  1. ORNL
  2. YAHSGS LLC, Richland, WA
Publication Date:
Research Org.:
Oak Ridge National Lab. (ORNL), Oak Ridge, TN (United States)
Sponsoring Org.:
Work for Others (WFO)
OSTI Identifier:
931473
DOE Contract Number:
DE-AC05-00OR22725
Resource Type:
Journal Article
Resource Relation:
Journal Name: Journal of Chemical Information and Modeling; Journal Volume: 47; Journal Issue: 2
Country of Publication:
United States
Language:
English
Subject:
97; 59 BASIC BIOLOGICAL SCIENCES; EVALUATION; FORECASTING; NEURAL NETWORKS; TOXICITY; METABOLISM; CALCULATION METHODS; neural network; informatics

Citation Formats

Piotrowski, Pamela L, Sumpter, Bobby G, Malling, Heinrich, Wassom, John, Lu, Po-Yung, Bothers, Robin, Sega, Gary, Martin, Sheryl A, and Parang, Morey. A Toxicity Evaluation and Predictive System Based on Neural Networks and Wavelets. United States: N. p., 2007. Web. doi:10.1021/ci6004788.
Piotrowski, Pamela L, Sumpter, Bobby G, Malling, Heinrich, Wassom, John, Lu, Po-Yung, Bothers, Robin, Sega, Gary, Martin, Sheryl A, & Parang, Morey. A Toxicity Evaluation and Predictive System Based on Neural Networks and Wavelets. United States. doi:10.1021/ci6004788.
Piotrowski, Pamela L, Sumpter, Bobby G, Malling, Heinrich, Wassom, John, Lu, Po-Yung, Bothers, Robin, Sega, Gary, Martin, Sheryl A, and Parang, Morey. Mon . "A Toxicity Evaluation and Predictive System Based on Neural Networks and Wavelets". United States. doi:10.1021/ci6004788.
@article{osti_931473,
title = {A Toxicity Evaluation and Predictive System Based on Neural Networks and Wavelets},
author = {Piotrowski, Pamela L and Sumpter, Bobby G and Malling, Heinrich and Wassom, John and Lu, Po-Yung and Bothers, Robin and Sega, Gary and Martin, Sheryl A and Parang, Morey},
abstractNote = {A computational approach has been developed for performing efficient and reasonably accurate toxicity evaluation and prediction. The approach is based on computational neural networks linked to modern computational chemistry and wavelet methods. In this paper we present details of this approach and results demonstrating its accuracy and flexibility for predicting diverse biological endpoints including metabolic processes, mode of action, and hepato- and neurotoxicity. The approach also can be used for automatic processing of microarray data to predict modes of action.},
doi = {10.1021/ci6004788},
journal = {Journal of Chemical Information and Modeling},
number = 2,
volume = 47,
place = {United States},
year = {Mon Jan 01 00:00:00 EST 2007},
month = {Mon Jan 01 00:00:00 EST 2007}
}
  • Quantitative structure-toxicity models were developed that directly link the molecular structures of a set of 50 alkylated and/or halogenated phenols with their polar narcosis toxicity, expressed as the negative logarithm of the IGC50 (50% growth inhibitory concentration) value in millimoles per liter. Regression analysis and fully connected, feed-forward neural networks were used to develop the models. Two neural network training algorithms (back-propagation and a quasi-Newton method) were employed. The best model was a quasi-Newton neural network that had a root-mean-square error of 0.070 log units for the 45 training set phenols and 0.069 log units for the five cross-validation setmore » phenols.« less
  • This paper presents a neural network based maximum power tracking controller for interconnected PV systems to commercial power sources. The neural network is utilized to identify the optimal operating voltage of the PV system. The controller generates the control signal in real time, and the control signal is fed back to the voltage control loop of the inverter to shift the terminal voltage of the PV system to the identified optimal one, which yields the maximum power generation. The controller is a PI type one. The proportion an the integral gains are set to their optimal values to achieve themore » fast response and also to prevent the overshoot and also the undershoot. The continuous measurement is required for the open circuit voltage on the monitoring cell, and also for the terminal voltage of the PV system. Because of the accurate identification of the optimal operating voltage of the PV system, more than 99% power is drawn for the actual maximum power.« less
  • In essentially all cases of intravenously administered administered radiolabeled antibody, hematologic toxicity has limited the amount of radioactivity that may be administered. In this study 4 potential predictors of hematologic toxicity are evaluated: red marrow dose (RMD), whole-body dose (WBD), administered activity (AA), and administered activity per unit surface area (AASA). These were evaluated against 8 measures of hematologic toxicity: grade (G), time-to-nadir (TTN), and absolute and per cent drop in counts (ADC, %DC) for platelets (P) and white blood cells (W). Data from a phase I study of 29 patients with metastatic colon carcinoma and a total of 35more » different antibody administrations of {sup 131}I-CC49 antibody were used in the analysis. Serum clearance kinetics were used to estimate RMD. A remainder-of-body term representing activity not in the red marrow was included in the RMD calculations. The whole-body dose was estimated as the product of whole-body cumulated activity, derived form survey meter measurements, and the whole-body to whole-body S-factor. All S-factors were obtained from MIRD Pamphlet 11. Patient surface area was estimated form each patient`s height and weight. Patients exhibiting accelerated clearance kinetics due to an immunologic response against the antibody (HAMA) were included in the analysis. The correlation coefficients, r, are listed in the table. All patients had been treated with 5-FU and leucovorin previously. When patients that has also been treated with mitomycin and/or alkylating agents were excluded from the analysis, the r values shown on the last line of the table were obtained for platelet toxicity grade. These results suggest that, although still poor, RMD may be a slightly better predictor of toxicity than other measures.« less
  • The results of a recent comprehensive evaluation of the relationship between four measures of in vitro genetic toxicity and the capacity of the chemicals to induce neoplasia in rodents carry some important implications. The results showed that while the Salmonella mutagenesis assay detected only about half of the carcinogenes as mutagens, the other three in vitro assays (mutagenesis in MOLY cells or induction of aberrations or SCEs in CHO cells) did not complement Salmonella since they failed to effectively discriminate between the carcinogens and noncarcinogens found negative in the Salmonella assay. The specificity of the Salmonella assay for this groupmore » of 73 chemicals was relatively high (only 4 of 29 noncarcinogens were positive). Therefore, the authors have begun to evaluate in vivo genetic toxicity assays for their ability to complement Salmonella in the identification of carcinogens.« less
  • Background. Hepatic arterial infusion (HAI) chemotherapy is effective for treating primary and metastatic carcinomas of the liver. Since hepatic arteries also supply the stomach and duodenum, HAI may result in unwanted infusion into the upper gastrointestinal tract and consequent gastric toxicity. Using fused images obtained with a combined SPECT/CT system, we assessed extrahepatic perfusion (EHP) and its correlation with gastrointestinal toxicity in patients receiving HAI. Methods. We studied 41 patients with primary or metastatic carcinoma of the liver who received HAI chemotherapy consisting of 5-fluorouracil and cisplatin. All underwent abdominal SPECT using a {sup 99m}Tc-MAA (185 MBq) instrument and anmore » injection rate of 0.1 ml/min, identical to the chemotherapy infusion rate. Delivery was through an implantable port. We analyzed the distribution of the anticancer agent on fused images and the relationship between EHP of the right gastric arterial region and gastric toxicity. All patients underwent esophagogastroduodenoscopy (EGDS). Results. Of the 41 patients, 11 (27%) manifested enhancement of the duodenal and gastric pyloric region on fused images. EGDS at the time of reservoir placement detected gastric ulcers in 10 of these patients. Conclusion. Fusion imaging with combined SPECT/CT reflects the actual distribution of the infused anticancer agents. The detection of EHP on fused images is predictive of the direct gastric toxicity from anticancer agents in patients undergoing HAI.« less