Discriminant analysis

Targeting bacterial growth in biofilm conditions: rational design of novel inhibitors to mitigate clinical and food contamination using QSAR

Antimicrobial resistance (AMR) is a pressing global issue exacerbated by the abuse of antibiotics and the formation of bacterial biofilms, which cause up to 80% of human bacterial infections. This study presents a computational strategy to address AMR by developing three novel quantitative structure-activity relationship (QSAR) models based on molecular topology to identify potential anti-biofilm and antibacterial agents. The models aim to determine the chemo-topological pattern of Gram (+) antibacterial, Gram (-) antibacterial, and biofilm formation inhibition activity. The models were applied to the virtual screening of a commercial chemical database, resulting in the selection of 58 compounds. Subsequent in vitro assays showed that three of these compounds exhibited the most promising antibacterial activity, with potential applications in enhancing food and medical device safety.

Beta-band power classification of go/no-go arm-reaching responses in the human hippocampus

Objective.Can we classify movement execution and inhibition from hippocampal oscillations during arm-reaching tasks? Traditionally associated with memory encoding, spatial navigation, and motor sequence consolidation, the hippocampus has come under scrutiny for its potential role in movement processing. Stereotactic electroencephalography (SEEG) has provided a unique opportunity to study the neurophysiology of the human hippocampus during motor tasks. In this study, we assess the accuracy of discriminant functions, in combination with principal component analysis (PCA), in classifying between 'Go' and 'No-go' trials in a Go/No-go arm-reaching task.Approach.Our approach centers on capturing the modulation of beta-band (13-30 Hz) power from multiple SEEG contacts in the hippocampus and minimizing the dimensional complexity of channels and frequency bins. This study utilizes SEEG data from the human hippocampus of 10 participants diagnosed with epilepsy. Spectral power was computed during a 'center-out' Go/No-go arm-reaching task, where participants reached or withheld their hand based on a colored cue. PCA was used to reduce data dimension and isolate the highest-variance components within the beta band. The Silhouette score was employed to measure the quality of clustering between 'Go' and 'No-go' trials. The accuracy of five different discriminant functions was evaluated using cross-validation.Main results.The Diagonal-Quadratic model performed best of the 5 classification models, exhibiting the lowest error rate in all participants (median: 9.91%, average: 14.67%). PCA showed that the first two principal components collectively accounted for 54.83% of the total variance explained on average across all participants, ranging from 36.92% to 81.25% among participants.Significance.This study shows that PCA paired with a Diagonal-Quadratic model can be an effective method for classifying between Go/No-go trials from beta-band power in the hippocampus during arm-reaching responses. This emphasizes the significance of hippocampal beta-power modulation in motor control, unveiling its potential implications for brain-computer interface applications.

Loss of LPAR6 and CAB39L dysregulates the basal-to-luminal urothelial differentiation program, contributing to bladder carcinogenesis

We describe a strategy that combines histologic and molecular mapping that permits interrogation of the chronology of changes associated with cancer development on a whole-organ scale. Using this approach, we present the sequence of alterations around RB1 in the development of bladder cancer. We show that RB1 is not involved in initial expansion of the preneoplastic clone. Instead, we found a set of contiguous genes that we term "forerunner" genes whose silencing is associated with the development of plaque-like field effects initiating carcinogenesis. Specifically, we identified five candidate forerunner genes (ITM2B, LPAR6, MLNR, CAB39L, and ARL11) mapping near RB1. Two of these genes, LPAR6 and CAB39L, are preferentially downregulated in the luminal and basal subtypes of bladder cancer, respectively. Their loss of function dysregulates urothelial differentiation, sensitizing the urothelium to N-butyl-N-(4-hydroxybutyl)nitrosamine-induced cancers, which recapitulate the luminal and basal subtypes of human bladder cancer.

MRI feature-based radiomics models to predict treatment outcome after stereotactic body radiotherapy for spinal metastases

OBJECTIVE

This study aimed to extract radiomics features from MRI using machine learning (ML) algorithms and integrate them with clinical features to build response prediction models for patients with spinal metastases undergoing stereotactic body radiotherapy (SBRT).

METHODS

Patients with spinal metastases who were treated using SBRT at our hospital between July 2018 and April 2023 were recruited. We assessed their response to treatment using the revised Response Evaluation Criteria in Solid Tumors (version 1.1). The lesions were categorized into progressive disease (PD) and non-PD groups. Radiomics features were extracted from T1-weighted image (T1WI), T2-weighted image (T2WI), and fat-suppression T2WI sequences. Feature selection involved intraclass correlation coefficients, minimal-redundancy-maximal-relevance, and least absolute shrinkage and selection operator methods. Thirteen ML algorithms were employed to construct the radiomics prediction models. Clinical, conventional imaging, and radiomics features were integrated to develop combined models. Model performance was evaluated using receiver operating characteristic (ROC) curve analysis, and the clinical value was assessed using decision curve analysis.

RESULTS

We included 194 patients with 142 (73.2%) lesions in the non-PD group and 52 (26.8%) in the PD group. Each region of interest generated 2264 features. The clinical model exhibited a moderate predictive value (area under the ROC curve, AUC = 0.733), while the radiomics models demonstrated better performance (AUC = 0.745-0.825). The combined model achieved the best performance (AUC = 0.828).

CONCLUSION

The MRI-based radiomics models exhibited valuable predictive capability for treatment outcomes in patients with spinal metastases undergoing SBRT.

CRITICAL RELEVANCE STATEMENT

Radiomics prediction models have the potential to contribute to clinical decision-making and improve the prognosis of patients with spinal metastases undergoing SBRT.

KEY POINTS

• Stereotactic body radiotherapy effectively delivers high doses of radiation to treat spinal metastases. • Accurate prediction of treatment outcomes has crucial clinical significance. • MRI-based radiomics models demonstrated good performance to predict treatment outcomes.

Machine learning approaches to predict drug efficacy and toxicity in oncology

In recent years, there has been a surge of interest in using machine learning algorithms (MLAs) in oncology, particularly for biomedical applications such as drug discovery, drug repurposing, diagnostics, clinical trial design, and pharmaceutical production. MLAs have the potential to provide valuable insights and predictions in these areas by representing both the disease state and the therapeutic agents used to treat it. To fully utilize the capabilities of MLAs in oncology, it is important to understand the fundamental concepts underlying these algorithms and how they can be applied to assess the efficacy and toxicity of therapeutics. In this perspective, we lay out approaches to represent both the disease state and the therapeutic agents used by MLAs to derive novel insights and make relevant predictions.

Rational Design of Chitin Deacetylase Inhibitors for Sustainable Agricultural Use Based on Molecular Topology

Fungicide resistance is a major concern in modern agriculture; therefore, there is a pressing demand to develop new, greener chemicals. Chitin is a major component of the fungal cell wall and a well-known elicitor of plant immunity. To overcome chitin recognition, fungal pathogens developed different strategies, with chitin deacetylase (CDA) activity being the most conserved. This enzyme is responsible for hydrolyzing the N-acetamido group in N-acetylglucosamine units of chitin to convert it to chitosan, a compound that can no longer be recognized by the plant. In previous works, we observed that treatments with CDA inhibitors, such as carboxylic acids, reduced the symptoms of cucurbit powdery mildew and induced rapid activation of chitin-triggered immunity, indicating that CDA could be an interesting target for fungicide development. In this work, we developed an in silico strategy based on QSAR (quantitative structure-activity relationship) and molecular topology (MT) to discover new, specific, and potent CAD inhibitors. Starting with the chemical structures of few carboxylic acids, with and without disease control activity, three predictive equations based on the MT paradigm were developed to identify a group of potential molecules. Their fungicidal activity was experimentally tested, and their specificity as CDA inhibitors was studied for the three best candidates by molecular docking simulations. To our knowledge, this is the first time that MT has been used for the identification of potential CDA inhibitors to be used against resistant powdery mildew strains. In this sense, we consider of special interest the discovery of molecules capable of stimulating the immune system of plants by triggering a defensive response against fungal species that are highly resistant to fungicides such as powdery mildew.

The origin of bladder cancer from mucosal field effects

Whole-organ mapping was used to study molecular changes in the evolution of bladder cancer from field effects. We identified more than 100 dysregulated pathways, involving immunity, differentiation, and transformation, as initiators of carcinogenesis. Dysregulation of interleukins signified the involvement of inflammation in the incipient phases of the process. An aberrant methylation/expression of multiple HOX genes signified dysregulation of the differentiation program. We identified three types of mutations based on their geographic distribution. The most common were mutations restricted to individual mucosal samples that targeted uroprogenitor cells. Two types of mutations were associated with clonal expansion and involved large areas of mucosa. The α mutations occurred at low frequencies while the β mutations increased in frequency with disease progression. Modeling revealed that bladder carcinogenesis spans 10-15 years and can be divided into dormant and progressive phases. The progressive phase lasted 1-2 years and was driven by β mutations.

QSAR study of phenolic compounds and their anti-DPPH radical activity by discriminant analysis

Phenolic compounds (PCs) could be applied to reduce reactive oxygen species (ROS) levels, and are used to prevent and treat diseases related to oxidative stress. QSAR study was applied to elucidate the relationship between the molecular descriptors and physicochemical properties of polyphenol analogues and their DPPH radical scavenging capability, to guide the design and discovery of highly-potent antioxidant substances more efficiently. PubMed database was used to collect 99 PCs with antioxidant activity, whereas, 105 negative PCs were found in ChEMBL database; their molecular descriptors were generated with Python's Rdkit package. While the molecular descriptors significantly related to the antioxidant activity of PCs were filtered by t-test. The prediction QSAR model was then established by discriminant analysis, and the obtained model was verified by the back-substitution and Leave-One-Out cross-validation methods along with heat map. It was revealed that the anti-DPPH radical activity of PCs was correlated with the drug-likeness and molecular fingerprints, physicochemical, topological, constitutional and electronic property. The established QSAR model could explicitly predict the antioxidant activity of polyphenols, thus were applicable to evaluate the potential of candidates as antioxidants.

Discriminant value of repetitive behaviors in families with autism spectrum disorder and obsessional compulsive disorder probands

Repetitive behaviors (RB) represent a wide spectrum of symptoms ranging from sensory-motor stereotypies to complex cognitive rituals, frequently dichotomized as low- and high-order sub-groups of symptoms. Even though these subgroups are considered as phenomenologically distinct in autism spectrum disorder (ASD) and obsessive-compulsive disorder (OCD), brain imaging and genetic studies suggest that they have common mechanisms and pathways. This discrepancy may be explained by the frequent intellectual disability reported in ASD, which blurs the RB expressivity. Given the high heritability of RB, that is, the diversity of symptoms expressed in the relatives are dependent on those expressed in their probands, we hypothesize that if RB expressed in ASD or OCD are two distinct entities, then the RB expressed in relatives will also reflect these two dimensions. We thus conduct a linear discriminant analysis on RB in both the relatives of probands with ASD and OCD and subjects from the general population (n = 1023). The discriminant analysis results in a classification of 81.1% of the controls (p < 10^-4 ), but poorly differentiated the ASD and OCD relatives (≈46%). The stepwise analysis reveals that five symptoms attributed to high-order RB and two related to low-order RB (including hypersensitivity) are the most discriminant. Our results support the idea that the difference of RB patterns in the relatives is mild compared with the distribution of symptoms in controls. Our findings reinforce the evidence of a common biological pattern of RB both in ASD and OCD but with minor differences, specific to each of these two neuro-developmental disorders. LAY SUMMARY: Repetitive behaviors (RB), a key symptom in the classification of both OCD and ASD, are phenomenologically considered as distinct in the two disorders, which is in contrast with brain imaging studies describing a common neural circuit. Intellectual disability, which is frequently associated with ASD, makes RB in ASD more difficult to understand as it affects the expression of the RB symptoms. To avoid this bias, we propose to consider the familial aggregation in ASD and OCD by exploring RB in the first-degree relatives of ASD and OCD. Our results highlight the existence of RB expressed in relatives compared to the general population, with a common pattern of symptoms in relatives of both ASD and OCD but also minor differences, specific to each of these two neuro-developmental disorders.

Macrolides May Prevent Severe Acute Respiratory Syndrome Coronavirus 2 Entry into Cells: A Quantitative Structure Activity Relationship Study and Experimental Validation

The global pandemic caused by the emerging severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) is threatening the health and economic systems worldwide. Despite the enormous efforts of scientists and clinicians around the world, there is still no drug or vaccine available worldwide for the treatment and prevention of the infection. A rapid strategy for the identification of new treatments is based on repurposing existing clinically approved drugs that show antiviral activity against SARS-CoV-2 infection. In this study, after developing a quantitative structure activity relationship analysis based on molecular topology, several macrolide antibiotics are identified as promising SARS-CoV-2 spike protein inhibitors. To confirm the in silico results, the best candidates were tested against two human coronaviruses (i.e., 229E-GFP and SARS-CoV-2) in cell culture. Time-of-addition experiments and a surrogate model of viral cell entry were used to identify the steps in the virus life cycle inhibited by the compounds. Infection experiments demonstrated that azithromycin, clarithromycin, and lexithromycin reduce the intracellular accumulation of viral RNA and virus spread as well as prevent virus-induced cell death, by inhibiting the SARS-CoV-2 entry into cells. Even though the three macrolide antibiotics display a narrow antiviral activity window against SARS-CoV-2, it may be of interest to further investigate their effect on the viral spike protein and their potential in combination therapies for the coronavirus disease 19 early stage of infection.

Computational analysis of macrolides as SARS-CoV-2 main protease inhibitors: a pattern recognition study based on molecular topology and validated by molecular docking

Macrolides share the same chemo-mathematical pattern as SARS-CoV-2 protease inhibitors.

Assessment of Luminal and Basal Phenotypes in Bladder Cancer

Genomic profiling studies have demonstrated that bladder cancer can be divided into two molecular subtypes referred to as luminal and basal with distinct clinical behaviors and sensitivities to frontline chemotherapy. We analyzed the mRNA expressions of signature luminal and basal genes in bladder cancer tumor samples from publicly available and MD Anderson Cancer Center cohorts. We developed a quantitative classifier referred to as basal to luminal transition (BLT) score which identified the molecular subtypes of bladder cancer with 80–94% sensitivity and 83–93% specificity. In order to facilitate molecular subtyping of bladder cancer in primary care centers, we analyzed the protein expressions of signature luminal (GATA3) and basal (KRT5/6) markers by immunohistochemistry, which identified molecular subtypes in over 80% of the cases. In conclusion, we provide a tool for assessment of molecular subtypes of bladder cancer in routine clinical practice.

Artificial intelligence with multi-functional machine learning platform development for better healthcare and precision medicine

Precision medicine is one of the recent and powerful developments in medical care, which has the potential to improve the traditional symptom-driven practice of medicine, allowing earlier interventions using advanced diagnostics and tailoring better and economically personalized treatments. Identifying the best pathway to personalized and population medicine involves the ability to analyze comprehensive patient information together with broader aspects to monitor and distinguish between sick and relatively healthy people, which will lead to a better understanding of biological indicators that can signal shifts in health. While the complexities of disease at the individual level have made it difficult to utilize healthcare information in clinical decision-making, some of the existing constraints have been greatly minimized by technological advancements. To implement effective precision medicine with enhanced ability to positively impact patient outcomes and provide real-time decision support, it is important to harness the power of electronic health records by integrating disparate data sources and discovering patient-specific patterns of disease progression. Useful analytic tools, technologies, databases, and approaches are required to augment networking and interoperability of clinical, laboratory and public health systems, as well as addressing ethical and social issues related to the privacy and protection of healthcare data with effective balance. Developing multifunctional machine learning platforms for clinical data extraction, aggregation, management and analysis can support clinicians by efficiently stratifying subjects to understand specific scenarios and optimize decision-making. Implementation of artificial intelligence in healthcare is a compelling vision that has the potential in leading to the significant improvements for achieving the goals of providing real-time, better personalized and population medicine at lower costs. In this study, we focused on analyzing and discussing various published artificial intelligence and machine learning solutions, approaches and perspectives, aiming to advance academic solutions in paving the way for a new data-centric era of discovery in healthcare.

A molecular risk score based on 4 functional pathways for advanced classical Hodgkin lymphoma

Despite improvement in the treatment of advanced classical Hodgkin lymphoma, approximately 30% of patients relapse or die as result of the disease. Current predictive systems, determined by clinical and analytical parameters, fail to identify these high-risk patients accurately. We took a multistep approach to design a quantitative reverse-transcription polymerase chain reaction assay to be applied to routine formalin-fixed paraffin-embedded samples, integrating genes expressed by the tumor cells and their microenvironment. The significance of 30 genes chosen on the basis of previously published data was evaluated in 282 samples (divided into estimation and validation sets) to build a molecular risk score to predict failure. Adequate reverse-transcription polymerase chain reaction profiles were obtained from 262 of 282 cases (92.9%). Best predictor genes were integrated into an 11-gene model, including 4 functional pathways (cell cycle, apoptosis, macrophage activation, and interferon regulatory factor 4) able to identify low- and high-risk patients with different rates of 5-year failure-free survival: 74% versus 44.1% in the estimation set (P < .001) and 67.5% versus 45.0% in the validation set (P = .022). This model can be combined with stage IV into a final predictive model able to identify a group of patients with very bad outcome (5-year failure-free survival probability, 25.2%).

Response diversity of Arabidopsis thaliana ecotypes in elevated [CO2] in the field

Free Air [CO(2)] Enrichment (FACE) allows for plant growth under fully open-air conditions of elevated [CO(2)] at concentrations expected to be reached by mid-century. We used Arabidopsis thaliana ecotypes Col-0, Cvi-0, and WS to analyze changes in gene expression and metabolite profiles of plants grown in "SoyFACE" (http://www.soyface.uiuc.edu/), a system of open-air rings within which [CO(2)] is elevated to approximately 550 ppm. Data from multiple rings, comparing plants in ambient air and elevated [CO(2)], were analyzed by mixed model ANOVA, linear discriminant analysis (LDA) and data-mining tools. In elevated [CO(2)], decreases in the expression of genes related to chloroplast functions characterized all lines but individual members of distinct multi-gene families were regulated differently between lines. Also, different strategies distinguished the lines with respect to the regulation of genes related to carbohydrate biosynthesis and partitioning, N-allocation and amino acid metabolism, cell wall biosynthesis, and hormone responses, irrespective of the plants' developmental status. Metabolite results paralleled reactions seen at the level of transcript expression. Evolutionary adaptation of species to their habitat and intrinsic genetic plasticity seem to determine the nature of responses to elevated [CO(2)]. Irrespective of their underlying genetic diversity, and evolutionary adaptation to different habitats, a small number of common, predominantly stress-responsive, signature transcripts appear to characterize responses of the Arabidopsis ecotypes in FACE.

Rabbit and nonhuman primate models of toxin-targeting human anthrax vaccines

The intentional use of Bacillus anthracis, the etiological agent of anthrax, as a bioterrorist weapon in late 2001 made our society acutely aware of the importance of developing, testing, and stockpiling adequate countermeasures against biological attacks. Biodefense vaccines are an important component of our arsenal to be used during a biological attack. However, most of the agents considered significant threats either have been eradicated or rarely infect humans alive today. As such, vaccine efficacy cannot be determined in human clinical trials but must be extrapolated from experimental animal models. This article reviews the efficacy and immunogenicity of human anthrax vaccines in well-defined animal models and the progress toward developing a rugged immunologic correlate of protection. The ongoing evaluation of human anthrax vaccines will be dependent on animal efficacy data in the absence of human efficacy data for licensure by the U.S. Food and Drug Administration.

Improved Diagnostic Classification of Alcohol Abusers by Combining Carbohydrate-deficient Transferrin and γ-Glutamyltransferase

Background: Biochemical markers can provide objective evidence of high alcohol consumption. However, currently available markers have limitations in their diagnostic performance.

Methods: The diagnostic values of the most frequently used markers [carbohydrate-deficient transferrin (CDT), γ-glutamyltransferase (GGT), aspartate aminotransferase, alanine aminotransferase, and mean corpuscular volume] were studied in an analysis of six different clinical studies (n = 1412) on alcohol abusers and social drinkers. The purpose of the analyses was to determine whether a combination of markers would improve the diagnosis of subjects.

Results: Discrimination between alcohol abusers and social drinkers, as measured by the areas under nonparametric ROC plots, was significantly better (P &lt;0.001) for the new combined marker [γ-CDT = 0.8 · ln(GGT) + 1.3 · ln(CDT)] than for any of the separate markers or combination of CDT or GGT with other markers. The cutoff values for γ-CDT (6.5) can be taken to be the same among males and females.

Conclusions: The combined variable γ-CDT is a powerful tool to discriminate alcohol abusers from social drinkers and is recommended for clinical use.

A new strategy to predict the neoplastic polyps of the gallbladder based on a scoring system using EUS

BACKGROUND

A new method to predict neoplastic polyps of the gallbladder using a scoring system based on five endoscopic ultrasonography (EUS) variables is presented.

METHODS

EUS data from patients with gallbladder polyps who were to undergo cholecystectomy were used for the construction of an EUS scoring system in polyps between 5 and 15 mm in diameter (reference group). The EUS scoring system developed from those patients was applied to other patients (validation group).

RESULTS

In the reference group, size was the most significant predictor of neoplastic polyp. All polyps 5 mm or less in diameter were non-neoplastic and 94% of polyps of greater than 15 mm were neoplastic in the reference group. For polyps between 5 and 15 mm in diameter, the area under the receiver-operating characteristic curves (ROC) plots for the endoscopic scoring system was significantly greater than that under the ROC plots for polyp size alone (p < 0.01). In the validation group, the risk of neoplastic polyp was significantly higher for polyps with a score of 6 or greater compared with those with a score of less than 6 (p < 0.01).

CONCLUSIONS

Our data show that a score based on five EUS variables identifies those patients at risk of neoplasia when polyps are between 5 and 15 mm in diameter. (Gastrointest Endosc 2000;52:372-9).

Proton magnetic resonance spectroscopy of cerebrospinal fluid in neurodegenerative disease: indication of glial energy impairment in Huntington chorea, but not Parkinson disease

Metabolite levels in cerebrospinal fluid from patients with Parkinson disease or Huntington chorea were compared with the levels in healthy controls using proton magnetic resonance spectroscopy. No significant differences were found for any metabolite measured in cerebrospinal fluid from patients with Parkinson disease compared to controls. Slight but significantly reduced levels of both lactate and citrate, however, were found in cerebrospinal fluid from patients with Huntington chorea compared to controls. This suggests possible impairment of both glycolysis and tricarboxylic acid cycle function. The reduction in lactate found in the present study may reflect neuronal loss. The decrease in citrate supports the theory of mitochondrial dysfunction in the brain of patients with Huntington chorea, but also suggests that there may be an important astrocytic component in this disease. If so, it would certainly have implications for neuronal function.

A multidisciplinary approach to toxicological screening: I. Systemic toxicity

The toxicity of 10 chemicals, including pesticides (carbaryl, chlordane, heptachlor, and triadimefon), solvents (carbon tetrachloride, dichloromethane, tetrachloroethylene, and trichloroethylene), and industrial chemicals [diethylhexylphthalate (DEHP) and phenol] was examined in the liver, kidneys, spleen, thymus, and adrenals of female F344 rats after 1 or 14 d of oral dosing. For each chemical, 4 doses were based on fractions of the acute LD50, which was estimated using an abbreviated (up-and-down) method. A multivariate analysis (MANOVA) was conducted for each organ using selected measures of toxicity. A post hoc contrast analysis was also conducted for significant MANOVA results in order to determine effective and ineffective doses. A single dose of heptachlor resulted in necrotic lymphocytes in the spleen and thymus at doses > or = 23 mg/kg. Triadimefon altered liver and spleen weights; this effect has not been described previously. Dichloromethane (> or = 337 mg/kg/d for 14 d) increased the incidence of necrosis of individual centrilobular hepatocytes. Trichloroethylene-induced hepatotoxicity was obtained at doses an order of magnitude lower than those reported in the literature. Acute DEHP (150 mg/kg) increased mitotic figures in hepatocytes, which were replaced by hepatocellular cytomegaly after 14 d of dosing at the same level. Following phenol exposure, there was an increased incidence in hepatocellular necrosis at 1 d, and an increased incidence of kidney lesions at 1 and 14 d; these findings were considered to be the result of vascular stasis. Overall, the algorithm used to select doses was effective for both 1- or 14-d dosing regimens. For all chemicals except carbon tetrachloride, the lowest effective dose for systemic toxicity was within the range of 3-56% of the LD50 for acute dosing, and 1-30% of the LD50 for repeated administration.

Neural network approach for computer-assisted interpretation of ultrasound images of the gallbladder

Multi-formatted films of 90 ultrasound examinations of the gallbladder (stones 56 cases, sludge 20 cases, hydrops five cases, normal gallbladder nine cases) have been digitalized and stored in a personal computer. Image data of each examination was processed to extract a 19-dimensional vector that represents the essential diagnostic information of each examination. This vector was evaluated by three different classification algorithms: (1) classical nearest neighbor principle, (2) classical linear discriminant analysis, (3) multilayered backpropagation neural network. The correct classification rate was 64% (58/90) for the nearest neighbor principle, 97% (87/90) for the linear discriminant analysis, and 99% (89/90) for the backpropagation neural network. We conclude that, (1) automated classification of ultrasound images is possible for limited diagnostic problems, (2) a neural network approach can be used successfully for that goal, and (3) the efficiency of the more flexible neural network approach is comparable to large-scale classical methods.

Anatomical risk factors for Kienböck's disease

Contralateral unaffected wrists from 41 males with Kienböck's disease were compared with wrists from 66 normal males. From X-rays, various features of the lunate and radius were measured. In patients with Kienböck's disease, the lunate tended to be smaller and inclined more radially than in normal subjects and the radial inclination was flatter. Discriminant analysis showed that 85% of the unaffected contralateral wrists in patients with Kienböck's disease and 74% of the wrists in normal subjects were accurately discriminated to their respective groups. It may be possible to identify subjects who are at risk for Kienböck's disease prior to onset using discriminant analysis.

Comparison of nonparametric recursive partitioning to parametric discriminant analyses in laboratory differentiation of hypercalcemia

Recursive partitioning analysis was compared to logistic, linear and quadratic discriminant analyses in the ability to differentiate hypercalcemic patients with primary hyperparathyroidism from those with malignancy. Stepwise discriminant analysis identified serum albumin as the best single discriminant test. Albumin decision values optimally separating the two hypercalcemic groups were 39.46, 38.54, and 32.25 g/l for the logistic, linear and quadratic discriminant methods, respectively. Recursive partitioning analysis identified carboxy-terminus parathyroid hormone (PTH) as the best discriminant test with an optimal decision value of 8.2 mequiv/l. The discrepancy between the selection of PTH by recursive partitioning analysis and albumin by discriminant techniques was attributed to the nonnormal distribution of PTH. Recursive partitioning analysis using PTH classified 85.4% of the patients correctly. Logistic, linear and quadratic methods, using albumin as the predictor variable, correctly classified 79.6%, 78.6%, and 79.6% of patients, respectively.

Short-term in vitro drug testing of mamma tumors: relation to estrogen and progesterone receptor levels, age, and proliferation rate

The correlation between age, proliferation rate of tumors, estrogen and progesterone receptors, and in vitro chemosensitivity to Adriamycin was studied on 43 primary mamma tumors. From an univariate statistical analysis of the results, it appeared that sensitive tumors, unlike resistant tumors, have fewer estrogen receptors and show a higher proliferation rate. And in addition, they are blocked to a greater extent by Adriamycin. In both groups age and progesterone receptors were not significantly different. A multivariate statistical analysis showed that in the classification into sensitive and resistant tumors, the percentage remaining incorporation after addition of Adriamycin and the proliferation rate contributed 94 and 5%, respectively. The first variable was the best measure for in vitro chemosensitivity. The classification of the tumors with the aid of a discriminant function proved to be successful in 91% of all the cases. No significant difference was observed between the in vitro sensitivity to Adriamycin when patients were divided into two groups according to age (less than or equal to 50 and greater than 50 years; 64 and 45% sensitive, respectively). This indicates that all patients benefit from the treatment. It also appeared that 85% of the estrogen negative tumors were sensitive to Adriamycin. So a chemotherapeutic instead of a hormonal therapy has to be considered for all ages, particularly in the case of estrogen negative receptors.

Tumour markers in the cerebrospinal fluid of patients with central nervous system metastases from extracranial malignancies

We have assessed the diagnostic value of the determination of cerebrospinal fluid lactate dehydrogenase, carcinoembryonic antigen, beta 2-microglobulin, beta-glucuronidase and total protein, using linear discriminant analysis, in detecting central nervous system metastases from extracranial malignancies. We conclude that, using these tests, it is impossible to differentiate between control individuals and patients with brain or epidural metastases. Leptomeningeal dissemination from either solid tumours or non-Hodgkin lymphoma could be differentiated from control individuals and patients with brain or epidural metastases. In this differentiation it is essential that bacterial, fungal or tuberculous meningitis be excluded from the differential diagnosis by other diagnostic procedures. The combination of beta-glucuronidase and beta 2-microglobulin provides almost the same diagnostic information as the combination of all parameters.

Serum cathodic trypsin-like immunoreactivity, pancreatic lipase, and pancreatic isoamylase as diagnostic tests of chronic pancreatitis or pancreatic steatorrhea

We compared serum concentrations of cathodic trypsin-like immunoreactivity, pancreatic lipase, and pancreatic isoamylase as diagnostic tests of chronic pancreatitis (and of pancreatic steatorrhea in the 41 patients with steatorrhea) in 105 patients (57 men, 48 women) consecutively investigated because of clinical suspicion of chronic pancreatitis. Chronic pancreatitis (36 patients), pancreatic steatorrhea (24 patients), and other diseases were diagnosed without knowledge of the serum levels of the three enzymes. When evaluated by means of receiver operating characteristic curves, no differences were found in diagnostic performance of the enzymes with regard to chronic pancreatitis or pancreatic steatorrhea. The sensitivity and specificity for recognition of chronic pancreatitis ranged from 0.306 to 0.444 and from 0.942 to 0.986 when the discrimination values were chosen to give highest efficiencies. The similar ranges for pancreatic steatorrhea were 0.500-0.708 and 0.882-0.941. In conclusion, none of the three enzymes had any advantage over the others as diagnostic tests of chronic pancreatitis or of pancreatic steatorrhea. Only positive test results have clinical importance because of the low sensitivities of the three enzymes.

Discriminant function analysis of clinical laboratory data. Use in alcohol research

There has been a growing interest in the use of discriminant function analysis of routinely collected laboratory data to differentiate between alcoholic and nonalcoholic individuals. However, no consensus has been reached concerning the validity and reliability of this form of analysis regarding the diagnosis of, or screening for, alcoholism. In addition, this technique may provide important information concerning the multivariate nature of the pathophysiological effects of alcohol. However, this issue is seldom discussed. Because of differences in methodology across studies and the growing confusion regarding conceptual issues associated with this statistical technique, there have been few critical reviews. This chapter is directed at critically evaluating the current literature, particularly with respect to the methodological and conceptual issues associated with the use of discriminant function analysis in alcohol research. As part of this approach we suggest guidelines for future research in this area.

Classification of malnutrition by statistical analysis of quantitative two-dimensional gel electrophoresis of plasma proteins

An attempt to use the relative concentrations of major plasma proteins for clinical assessment of severe malnutrition is described. Quantitative two-dimensional gel electrophoresis was used to measure the concentrations of 24 major proteins in small aliquots of plasma obtained from children, aged 0 to 3 years, who were patients and outpatients in Liberian hospitals. Fifteen had a clinical diagnosis of kwashiorkor, 36 were diagnosed with marasmus, and 18 were controls. There were also 5 controls from the United States. The individuals were placed in six groups; kwashiorkor, kwashiorkor who died during treatment, marasmus, marasmus who died, Liberian controls, and U.S. controls. The amount of protein in each spot in the two-dimensional gels was estimated by measuring bound stain using a laser scanner and computerized image analysis. We found very low serum transferrin levels in malnourishment, in agreement with reports from other investigators. All of the data for 24 protein variables were pooled for factor analysis; the mean factor scores for each group differed, with the kwashiorkor groups furthest from the controls. Results of discriminant analysis using the amounts of different numbers of protein variables (3 to 24) were compared for posterior assignment of individuals to groups. The validity of the method was tested by analysis of plasma aliquots obtained from patients following initiation of therapy and which were not a part of the training set. Predictive performance (prognosis of patient survival) depended upon the number of protein variables used. Although artifactual fitting of the data is expected to contribute to performance as the number of variables is increased, use of as many as 7 variables may be justified, even with our small patient groups. Possible use of these results for development of a practical clinical test is discussed.

Evaluation and comparison of cathodic trypsin-like immunoreactivity, pancreatic lipase and pancreatic isoamylase in the diagnosis of acute pancreatitis in 849 consecutive patients with acute abdominal pain

In 849 patients (417 men, 432 women) consecutively hospitalized with acute abdominal pain we compared the value of serum cathodic trypsin-like immunoreactivity, pancreatic lipase (EC 3.2.1.3) and pancreatic isoamylase (EC 3.2.1.1) as diagnostic tests for acute pancreatitis. The diagnoses of acute pancreatitis (in 49 patients, 5.8%) and other diseases were made without knowledge of these enzyme values. When evaluated by means of receiver operating characteristic curves no differences were found in diagnostic performance of the three enzymes. Use of combinations of different enzymes had no advantage over single enzyme determination using discrimination analysis for evaluation. The highest efficiency was for all three enzymes 0.991 (95% confidence limits: 0.983-0.995) and for all three enzymes the discrimination value giving this efficiency was several times the upper limit of reference range: 1 779 micrograms/l for cathodic trypsin-like immunoreactivity, 831 U/l for pancreatic isoamylase and 316 micrograms/l for pancreatic lipase. None of the enzymes had any prognostic value at admission in predicting a mild or severe attack of acute pancreatitis. In conclusion, no single enzyme or combination of enzymes had any diagnostic advantage for acute pancreatitis in patients with acute abdominal pain. Thus selection of one of the three enzymes as diagnostic test of acute pancreatitis is to be based on considerations such as economy, methodological simplicity, possibility of automated assay and the time-consumption at the assay.

Follow-up and prospective studies of the classification of liver disease

Two hundred patients with different liver diseases were observed during a period of 6-8 years. The diagnosis at the first hospitalization was based on morphological criteria (and, in some cases, additional clinical information). In 162 of the cases an initial 'specific' diagnosis could be made. By the time of the follow-up study the diagnosis was confirmed in 73% of them. In 22 of 38 patients who were initially unclassifiable, the diagnosis was made definite by the follow-up study. Eighty-five patients were hospitalized for re-examination 6-8 years after the initial study. Several of the liver diseases initially had quite typical patterns of clinical chemical data. Allocation by discriminant analysis was therefore in good agreement with the morphological classification. The follow-up study showed that several patients with initially atypical patterns of clinical chemical results had their diagnosis changed. In 35 patients with the final diagnosis of chronic active hepatitis (CAH) or primary biliary cirrhosis (PBC), laboratory data from the last hospitalization were used for discriminant analysis with teaching data from the initial study. Ninety-seven per cent were correctly allocated, and we conclude that these patients retain recognizable patterns of laboratory results for several years, even when given immunosuppressive treatment. The potential clinical usefulness of discriminant analysis of laboratory data for differential diagnosis was evaluated by a prospective study of 65 patients with the morphological diagnosis of CAH or PBC. Correspondence with the morphological classification system was found in almost 90% of the cases.

Use of a microcomputer for the definition of multivariate confidence regions in medical diagnosis based on clinical laboratory profiles

The use of multivariate confidence regions is proposed for the classification of clinical laboratory profiles into diagnostic classes. For this purpose, a multivariate confidence region is developed for each diagnostic class. Three methods (UNEQ, EQ, and SIMCA) are evaluated and compared with classical linear discriminant analysis. As an example, a small data set concerning the differentiation of the thyroid functional states on the basis of five laboratory tests is used. It is shown that related procedures can produce results of very different quality and that the multivariate region approach is attractive for the clinician's daily practice since the methods are easily implemented on a microcomputer.

A multivariate statistical method for the establishment of maximum allowable exposure to toxic materials in the workplace

A model has been developed for setting standard limiting concentrations for toxic substances in the body fluid or tissues of industrial workers. The method is very flexible, and allows the use of either univariate or multivariate data, and of classification schemes which are either supervised, unsupervised, or partially supervised. An index of subclinical biological damage is derived which is used to determine an optimal limiting exposure level. The model is applied to the particular case of selecting a maximum allowable concentration of blood lead in workers in industrial settings where lead toxicity is a danger.

Computerized interpretation of laboratory tests: an overview of systems, basic principles and logic techniques

The ever-increasing volume and complexity of laboratory testing demands that interpretive reporting schemes be initiated. However, because of the massive volume of laboratory data, computers must be employed in this task. To design such programs, a clear understanding of the diagnostic process is required. There are numerous factors which must be respected for reports to be designed in an optimal fashion. Reports can vary in complexity from simple normal range listings to elaborate mathematical modelling of physiologic processes. A variety of well-studied logic techniques can be employed. These features have been studied during the development of the LABDOC computerized reporting system.

Differential diagnostic value in hepatobiliary disease of serum conjugated bile acid concentrations and some routine liver tests assessed by discriminant analysis

Serum concentrations of glycine and taurine conjugates of cholic, chenodeoxycholic, and deoxycholic acid were measured with a HPLC-enzymatic assay in 104 patients with hepatobiliary diseases. The capability of discriminating between four major diagnostic categories by the serum bile acid concentrations, alone and in conjunction with five commonly used biochemical liver tests, was evaluated by stepwise linear discriminant analysis. The serum bile acid pattern alone was inferior to the routine liver tests in separating the diagnostic groups. When the two sets of analytes were combined, the generalized group distance (Rao's V) was significantly increased, showing that the serum bile acid pattern contained discriminatory information in addition to the routine liver tests. The gain in correct reclassification of patients obtained by adding the serum bile acids, however, was only marginal.

Free thyroxine, thyroxine/TBG ratio and other in vitro tests of thyroid function evaluated by discriminant analysis

T4, T3, TSH, FT4 (Corning method), TBG, T3U and a diversity of free thyroid hormone indices were determined in serum from 72 reference persons and 337 patients with either thyroid diseases, TBG abnormalities, non-thyroid illnesses, or medications known to affect the results of thyroid function tests. Ninety-six per cent were 20 years or older. None of the laboratory parameters showed significant age or sex dependence, whereas systemic illness and medications frequently influenced test results. Thus the mean values of all patient groups differed significantly from those of the reference group in two or more laboratory tests, and significantly from those of the reference group in two or more laboratory tests, and significant differences were detected in 73% of the total number of 209 paired comparisons of test results between reference and patient groups. As a single parameter, free thyroxine index was found to be a marginally more efficient thyroid function discriminator than FT4, which again was more efficient than the T4/TBG ratio. However, non of these parameters were included in the combinations of laboratory tests identified as optimum by discriminant analysis. The overall optimum test combination was T4, TSH and T3U, which had an efficiency for correct thyroid function reclassification better than 98% in the total material. T3 replaced TSH in distinguishing between hyperthyroidism and euthyroidism.

Screening for thyroid disorders in middle-aged women by computer-assisted evaluation of a thyroid hormone panel

Free thyroxine index, free tri-iodothyronine index and thyrotropin were determined in connection with gynecological screening for cervical carcinoma on middle-age women. A total of 3885 women were investigated. Women with values outside a trivariate reference region were reinvestigated by renewed hormone determinations and those with hormone results still outside the reference region were subjected to clinical and further laboratory investigations. Thus 20 women (0.51% of the study population) with previously unknown hyperthyroidism and 25 women (0.64%) with previously unknown hypothyroidism were identified and successfully treated. Values outside the reference region were also studied by cluster analysis, which separated hyper- and hypo thyroid patients into different clusters. but allowed no further identification of different thyroid disorders. We conclude that laboratory screening for thyroid disorders may be justified in middle aged women if performed together with another screening program. Computer-assisted multivariate evaluation of the hormone results was of value of identification of subjects with thyroid disorders but their numerical classification by cluster analysis was of little practical value.

Initiatives toward effective decision making and laboratory use

Escalating health care costs constitute a public issue of paramount importance today, Among the leading growth factors in this rise is the cost of hospital services, notably laboratory services. With respect to the clinical laboratory, rising costs appear to be almost entirely attributable to expanding utilization and introduction of new services. The clinical laboratory has gone through a technological revolution in two decades that has changed it from a largely manual to a highly automated system of great speed and capacity. This change had produced a change in the style of providing services, a change that includes the provision of quantities of unsolicited data. A parallel change in the style of use of the laboratory has taken place on the part of patient care physicians from a relatively sparing, problem oriented use pattern to a relatively lavish, data oriented one. These reciprocal changes have transformed medicine, in the United States, at least, into a relatively high laboratory use culture. Abandonment of the new technology and return to a simpler, more primitive laboratory world would be a drastic and most inappropriate response to the new situation. Furthermore, arbitrary measures such as rationing, quotas, and tariffs are, if enacted, almost certain to fail. The most effective long term strategies, though more demanding of time and effort, lie through modification of physician behavior through the pathways of education and research. Education and research initiatives now in progress can in time influence laboratory use patterns of physicians at all career levels, improving the logic of test use and providing more strategic, prudent, and cost effective overall laboratory utilization practices. These approaches will require much improved communication between laboratory and bedside and a new intense involvement of laboratory physicians and scientists in the tasks of helping to improve the use of laboratory tests and laboratory data.