Showing papers on "False positive paradox published in 1998"

Intrusion detection using sequences of system calls

Abstract: A method is introduced for detecting intrusions at the level of privileged processes. Evidence is given that short sequences of system calls executed by running processes are a good discriminator between normal and abnormal operating characteristics of several common UNIX programs. Normal behavior is collected in two waysc Synthetically, by exercising as many normal modes of usage of a program as possible, and in a live user environment by tracing the actual execution of the program. In the former case several types of intrusive behavior were studieds in the latter case, results were analyzed for false positives.

Hidden Markov models for detecting remote protein homologies.

Abstract: MOTIVATION A new hidden Markov model method (SAM-T98) for finding remote homologs of protein sequences is described and evaluated. The method begins with a single target sequence and iteratively builds a hidden Markov model (HMM) from the sequence and homologs found using the HMM for database search. SAM-T98 is also used to construct model libraries automatically from sequences in structural databases. METHODS We evaluate the SAM-T98 method with four datasets. Three of the test sets are fold-recognition tests, where the correct answers are determined by structural similarity. The fourth uses a curated database. The method is compared against WU-BLASTP and against DOUBLE-BLAST, a two-step method similar to ISS, but using BLAST instead of FASTA. RESULTS SAM-T98 had the fewest errors in all tests-dramatically so for the fold-recognition tests. At the minimum-error point on the SCOP (Structural Classification of Proteins)-domains test, SAM-T98 got 880 true positives and 68 false positives, DOUBLE-BLAST got 533 true positives with 71 false positives, and WU-BLASTP got 353 true positives with 24 false positives. The method is optimized to recognize superfamilies, and would require parameter adjustment to be used to find family or fold relationships. One key to the performance of the HMM method is a new score-normalization technique that compares the score to the score with a reversed model rather than to a uniform null model. AVAILABILITY A World Wide Web server, as well as information on obtaining the Sequence Alignment and Modeling (SAM) software suite, can be found at http://www.cse.ucsc.edu/research/compbi o/ CONTACT karplus@cse.ucsc.edu; http://www.cse.ucsc.edu/karplus

False positives, false negatives, and the validity of the diagnosis of major depression in primary care

Abstract: Objective To explore the issues of diagnostic specificity and psychiatric "caseness" (i.e., whether a patient meets the conditions to qualify as a "case" of a disease or syndrome) for major depression in the primary care setting. Design A cross-sectional study comparing the demographic, clinical, and mental health characteristics of patients identified as depressed by their family physicians with those meeting diagnostic criteria for major depression on the criterion standard Structured Clinical Interview for Diagnostic and Statistical Manual of Mental Disorders, Revised Third Edition. Setting The offices of 50 family physicians from private and academic practice in southeast Michigan. Patients A total of 1580 consecutive adult patients being seen for routine primary care services, from whom a weighted sample of 372 patients completed a set of mental health screening and diagnostic instruments. Main outcome measures Patients were assigned to 1 of 4 groups (true positive, false positive, false negative, and true negative) based on clinician identification and Structured Clinical Interview for Diagnostic and Statistical Manual of Mental Disorders, Revised Third Edition diagnosis. Differences between the 4 groups in demographic and clinical characteristics, scores on mental health instruments and mental health history were explored. Results Physician identification of depression was strongly associated with increased familiarity with the patient and the presence of suggestive clinical cues, such as history of or treatment for depression, patient distress, and presence of vegetative symptoms. Patients in the false-positive group displayed significantly higher levels of distress and impairment and were significantly more likely to have a history of mental health problems and treatment than were those in the true-negative group. The 2 "misidentified" groups, false positives and false negatives, were indistinguishable in their clinical characteristics (impairment, distress, or mental health history). Both groups' scores occupied the middle ground between true positives and true negatives on most clinical characteristics. Physicians appeared to discriminate between these 2 groups on the basis of their knowledge of the patient's clinical history. Conclusions Misidentification of depression in primary care may be in part an artifact of the use of the psychiatric model of caseness in the primary care setting. Our results are most consistent with a chronic disease-based model of depressive disorder, in which patients classified as false positive and false negative occupy a clinical middle ground between clearly depressed and clearly nondepressed patients. Family physicians appear to respond to meaningful clinical cues in assigning the diagnosis of depression to these distressed and impaired patients.

CAT scans, PET scans, and genomic scans.

Abstract: As we begin the long march toward genetic dissection of complex traits, it becomes necessary to develop optimum study designs and retool ourselves to face the emerging new challenges. Key issues pertaining to the design of genomic scans are reviewed, including: sampling unit, definition and refinement of phenotype, genotyping issues, one-stage vs. two-stage strategies, sample size and power, and cost and feasibility. It is emphasized that false positives should not be minimized in isolation from the issue of false negatives. Striking a practical balance between the two error rates is suggested. In terms of future directions to pursue, three areas are suggested: meta-analysis for pooling linkage results from multiple scans, rapid multivariate screening methods for increased power to detect quantitative trait loci (QTLs), and classification and regression trees (CART) methodology for handling heterogeneity and interactions. Finally, three recommendations are proposed for genomic scans. First, so as to minimize false negatives for a fixed sample size, it is recommended that we tolerate/accept a reasonable rate of false positives, on average, one false positive per individual scan. Second, so as to enable the use of relatively strict significance levels for interpreting the results from a genomic scan, it is highly recommended that the sample size be derived based on a significance level of at most 0.01 (and not 0.05) and 90% power (and not 80%). Third, it is recommended that the stringent significance levels suggested by Lander and Kruglyak be used when pooling evidence from multiple genomic scans (and not at the level of individual scans).

Analysis of methods for reducing false positives in the automated detection of clustered microcalcifications in mammograms

Abstract: Clustered microcalcifications are often the first sign of breast cancer in a mammogram. Nevertheless, all clustered microcalcifications are not found by an individual radiologist reading a mammogram. The use of a second reader may find those clusters of microcalcifications not found by the first reader, thereby improving the sensitivity of detecting clustered microcalcifications. Our laboratory has developed a computerized scheme for the detection of clustered microcalcifications, which can act like a second reader, that is undergoing clinical evaluation. This paper concerns the feature analysis stage of the computer scheme, which is designed to remove some of the false-computer detections. We have examined three methods of feature analysis, namely, rule based (the method currently used), an artificial neural network (ANN), and a combined method. In an independent database of 50 images, at a sensitivity of 83%, the average number of false positive (FP) detections per image was: 1.9 for rule-based, 1.6 for ANN, and 0.8 for the combined method. We demonstrate that the combined method performs best because each of the two stages eliminates different types of false positives.

Elimination of false positives generated through PCR re-amplification of differential display cDNA

Abstract: Differential display (DD) is a powerful molecular tool that allows the identification and subsequent isolation of transcripts differentially expressed between biological samples, for example, between undifferentiated and differentiated cells, between different tissues or in one tissue at different stages of development. However, significantly high rates of apparent false positives have been reported using this technique. We suggest that the vast majority of false positives do not represent the originally selected transcript, but instead result from the re-amplification of cDNA species that co-migrate with the cDNA of interest in DD gels. Here we describe the use of a procedure to resolve co-migrating cDNAs and to purify the candidate of interest before cloning. The use of this modified technique resolves downstream problems encountered during DD experiments.

Computer-aided diagnoses: Automatic detection of lung nodules

Abstract: This work describes a computational scheme for automatic detection of suspected lung nodules in a chest radiograph. A knowledge-based system extracts the lung masks over which we will apply the nodule detection process. First we obtain the normalized cross-correlation image. Next we detect suspicious regions by assuming a threshold. We examine the suspicious regions using a variable threshold which results in the growth of the suspicious areas and an increase in false positives. We reduce the large number of false positives by applying the facet model to the suspicious regions of the image. An algorithmic classification process gives a confidence factor that a suspicious region is a nodule. Five chest images containing 30 known nodules were used as a training set. We evaluated the system by analyzing 30 chest images with 40 confirmed nodules of varying contrast and size located in various parts of the lungs. The system detected 100% of the nodules with a mean of six false positives per image. The accuracy and specificity were 96%.

Efficient Focusing and Face Detection

Abstract: We present an algorithm for shape detection and apply it to frontal views of faces in still grey level images with arbitrary backgrounds. Detection is done in two stages: (i) “focusing,” during which a relatively small number of regions-of-interest are identified, minimizing computation and false negatives at the (temporary) expense of false positives; and (ii) “intensive classification,” during which a selected region-of-interest is labeled face or background based on multiple decision trees and normalized data. In contrast to most detection algorithms, the processing is then very highly concentrated in the regions near faces and near false positives.

The Association of False-Positive Rapid Plasma Reagin Results and HIV Infection

Abstract: Background and Objectives: To study the relationship between a false positive rapid plasma reagin (RPR) result (FP), syphilis, and HIV infection in our patients. Methods: A prospective study of the incidence of FP tests and syphilis in the general population and its relationship to HIV infection over a period of 6 months. Results: 8.76% of the population were HIV positive. False positives were found in 15% and 1.2% of the HIV infected and noninfected patients, respectively; the attributable risk for HIV was 14.97. Syphilis was found in 5% and 0.9% of the positive and negative HIV patients, respectively; the attributable risk for HIV was 5.4. Conclusions: The incidence of false positive RPR results in the HIV-infected population is significantly higher than that of the non-HIV-infected patients.

Parsing TV programs for identification and removal of nonstory segments

Abstract: ing video information automatically from TV broadcast, requires reliable methods for isolating program and commercial segments out of the full broadcast material. In this paper, we present the results from cut, static sequence, black frame, and text detection, for the purpose of isolating non-program segments. These results are evaluated, by comparison, to human visual inspection using more than 13 hours of varied program content. Using cut rate detection alone, produced a high recall with medium precision. Text detection was performed on the commercials, and the false positive segments. Adding text detection slightly lowers the recall. However, much higher precision is achieved. A new fast black frame detector algorithm is presented. Black frame detection is important for identifying commercial boundaries. Results indicate that adding detection of text, in addition to cut rate, to reduce the number of false positives, appears to be a promising method. Furthermore, by adding the information about position and size of text, and tracking it through an area, should further increase reliability.

The esophageal detector device can give false positives for tracheal intubation

Abstract: The esophageal detector device is a useful, inexpensive, portable device for the detection of inadvertent intubation of the esophagus. It has been extensively tested. It is very accurate with occasional false negatives for tracheal intubation but only two reported cases of false positives for tracheal intubation previously described. This article presents two further cases of false positives for tracheal intubation using the esophageal detector device and warns that although it is the more accurate of the cheap and portable devices available, it is not perfect.

Automatic detection of microcalcifications in digital mammography

Abstract: We devised, built and tested a detector and segmenter of microcalcifications in mammograms. Our segmenter includes preprocessing, extraction of 17 features, genetic selection of the best subset of six features, and a k-nearest neighbor classifier to suppress false candidates. From regions of interest in 86 clinically proven digital mammograms (one region per mammogram), we constructed a set of 716 manually determined microcalcifications and a set of 1222 false candidates. Our algorithm detected 695 (97%) and missed 21 (3%) of the 716 microcalcifications and produced 220 (18%) false positives among the 1222 false candidates. We evaluated this automated segmenter by comparing its diagnostic efficiency to that of a manual segmenter. All of the cases represented by the 86 regions of interest in our test were biopsied in earlier examinations. Among these 57 were proved benign. The diagnostic system containing the manual segmenter reduced this number to 16 (a 72-percent reduction). The diagnostic system containing our automated segmenter produced 20 benign biopsies (a 68-percent reduction). The loss of these four percentage points (from 72 to 68) is the price paid for elimination of the tedious and costly task of manual segmentation, which suffers from the variability and fatigue associated with human perception.

The Delayed Memory Controversy in an Epidemiological Framework

Abstract: Epidemiological concerns about base rates, true and false positives, and the possibility of an epidemic of therapy-induced false memories are frequently raised in the debate over delayed memories of childhood sexual abuse (CSA). Most references to epidemiological concepts emphasize the potential problem of false positives arising from therapists' assessments of CSA but neglect the potential problem of false negatives. A more balanced epidemiological analysis that includes estimations of the risks of both false positives and false negatives is more relevant to clinical concerns and better identifies gaps in our understanding of this area. This article presents estimates for certain epidemiological variables based on available research and a range of suppositions. The evidence at hand suggests that the rate of false negatives for CSA history poses a larger epidemiological problem than the rate of false positives.

Benefits and limitations of mass screening. The natural history of breast cancer

Abstract: Most mortalities caused by breast cancers are due to metastases. A breast cancer gives rise to lymphatic spread, which is at the origin of nodal involvement, and to distant dissemination, usually through the bloodstream. Nodal involvement, which generally occurs before metastatic dissemination, appears not to be the cause of the dissemination but rather an index of the likelihood of a tumor to metastasize. This likelihood is influenced by the histological grade and the growth rate of the tumor. Despite the variations in biological characteristics of breast tumors and the likelihood of metastatic dissemination, the existence of a relationship between tumor size and the probability of metastatic dissemination enables the calculation of how many disseminations could be avoided through early diagnosis and thus the gain in human life that screening could hope to achieve. However, in reality, the effective gain is lesser because the rate of participation is lower than 100% and the existence of false negatives (undetected cancers) reduces the number of detected cancers. Furthermore, a high rate of false positives increases the anxiety of women because they provoke unnecessary examinations. Screening is worthwhile only if the increase in human life outweighs the economic and social costs (anxiety, going to appointments) that it may produce. It is therefore necessary to improve the benefits (fewer false negatives) and to decrease the social and psychological costs (fewer false positives). This can be done by the implementation of rigorous quality assurance, systematic training of health care personnel, a follow-up of women who have been screened, and an annual assessment of screening results. Screening in France since 1994 has shown a marked improvement since the implementation of guidelines prepared by the national screening committee which emphasize the need for quality control and ongoing training of staff. However, the results vary greatly from one county to another and are often below European standards, although some regions in France are close to achieving these. This shows that the system of screening in France, despite its organization involving many private and public radiology departments, can give results equal to those in countries in which screening is performed by a small number of specialized units. The French system requires regional structures of orientation and evaluation. To be effective, these structures must be guided by written legislation and regulations, otherwise they will be unable to overcome the difficulties currently faced by the implementation of mass screening.

Automated Detection of Clustered Microcalcifications on Digitized Mammograms

Abstract: We have been developing an automated detection scheme for clustered microcalcifications on digital mammograms and reported the methods in several papers [1]–[4]. These schemes show a good performance in detection, but there is a problem that many false-positive candidates (ten and more) appear in some specific images. Therefore, an improvement of the elimination step of false positives is required. To achieve this, we have developed new methods of discrimination of the candidates and elimination of the false positives.

Model-Based Data Fusion and Discrimination of UXO in Magnetometry and EM Surveys

Abstract: : Locating, identifying and disposing of buried UXO on the 10 million acres of contaminated lands in the continental United States is a 500 billion dollar problem. Development of new technologies with improved data analysis has been identified as a high priority triservice requirement. Using current methods, it has been shown that false alarm detections far outnumber correctly identified ordnance. The best performing technologies typically have a false alarm rate of 300-500%. The high cost of digging and siposing of targets accounts for the overwhelming portion of the costs of UXO remediation, therefore a substantial saving could be recognized if the number of false positives were reduced. Using data collected by the Naval Research Laboratory's MTADS, new software techniques are being developed to improve discrimination and reduce the false alarm rate.

Characterization of small inhomogeneities

Abstract: X-ray mammography, yet very useful, has limitations. Even with considerable training, it produces false negatives and false positives.

[Tri-test: clinical considerations on 1784 cases].

Abstract: Background To evaluate the triple screen serum test as a noninvasive screening test for expectant mothers > 35 years old, who are not usually considered for invasive screening for trisomy 21. Methods 1784 tri-tests (triple serum screening tests) were performed on expectant mothers between their 15th and 18th week of pregnancy, using the radioimmunological Ria-Kodak with an Alpha program (Logical Medical System LTD) with a cut-off value of 1:350. Results 244 positive equal to 13.60%. The percentage of false positives was respectively 12.9% (age 35). Only in two of these cases did we have a positive response in amniocentesis for a fetus affected with Down syndrome. Of the 1540 patients with a negative tri-test, one woman gave birth to a fetus with Down syndrome. Conclusions Our study revealed a sensitivity of 66%. The elevated number of false positives has led us to decide on a variation on the cut off in the future: from 1:350 to 1:300.

Method and system for automated multi-sampled detection of lesions in images

Abstract: Method and system for automated detection of lesions in medical images using multi-sampling. An image is obtained and digitized. A multi-sampling (30) technique is used to generate a plurality of sampled images by sampling the digital image a corresponding plurality of times. The sampled images will have subtle differences from each other. All or a selected number of the sampled images are subjected to a lesion detection routine to detect regions suspected of being a lesion. The results of the detection are compared and regions that appear in less than a predetermined number of the sampled images are determined to be a false positive and eliminated. The remaining regions are determined to be a true positive. The detected lesions (13) could then undergo further analysis such as feature extraction (36) and feature analysis. Using the method and system according to the invention, the number of false positives can be reduced, improving the sensitivity of the automated detection.

On clone assembly problems: an error-tolerant test for interval graphs

Abstract: An important problem in DNA physical mapping is to reassemble the clone fragments to determine the structure of the entire molecule. The error-free version of this problem can be modeled as an interval graph recognition problem, where an interval graph is the intersection graph of a collection of intervals. However, since the data collected from laboratories almost surely contain some errors, traditional recognition algorithms can hardly be applied directly. We present a new test which has the following features: 1) the algorithm assembles the clones efficiently when the data is error-free; 2) in a case when the error rate is small (say, less than 3%) the test can likely detect and automatically correct the following three types of errors false positives, false negatives and chimeric clones; and 3) the test also identifies those parts of the data that are problematic, thus allowing biologists to perform further experiments to clean up the data.