Machine learning for predicting chronic diseases: a systematic review.

OBJECTIVES: We aimed to review the literature regarding the use of machine learning to predict chronic diseases. STUDY DESIGN: This was a systematic review. METHODS: The searches included five databases. We included studies that evaluated the prediction of chronic diseases using machine learning models and reported the area under the receiver operating characteristic curve values. The Transparent Reporting of a multivariable prediction model for Individual Prognosis Or Diagnosis scale was used to assess the quality of studies. RESULTS: In total, 42 studies were selected. The best reported area under the receiver operating characteristic curve value was 1, whereas the worst was 0.74. K-nearest neighbors, Naive Bayes, deep neural networks, and random forest were the machine learning models most frequently used for achieving the best performance. CONCLUSION: We found that machine learning can predict the occurrence of individual chronic diseases, progression, and their determinants and in many contexts. The findings are original and relevant to improve clinical decisions and the organization of health care facilities.

In vitro hepatotoxicity of SR 4233 (3-amino-1,2,4-benzotriazine-1,4-dioxide), a hypoxic cytotoxin and potential antitumor agent.

SR 4233 (3-amino-1,2,4-benzotriazine-1,4-dioxide) is presently undergoing investigation as an antitumor agent because of its high selective toxicity for hypoxic cells in vitro and in vivo. It has been found to be 15 to 200 times more toxic to hypoxic rodent and human cell lines than their normoxic counterparts. We investigated the toxicity of SR 4233 in primary cultures of hepatocytes under various oxygen tensions, ranging from 1% to 20% oxygen. The 50% lethal dose of SR 4233 was found to be 50 times lower in hepatocyte monolayers at 1% O2 versus 20% O2. Even at 4% O2, a concentration that prevails in the pericentral area of the liver under conditions of normal blood flow, SR 4233 was an order of magnitude more toxic than at 20% O2. All samples were analyzed for metabolites, and metabolism was found to be dependent on both the SR 4233 concentration and the oxygen tension. Formation of the major metabolite SR 4317 occurred to the greatest extent at the lowest oxygen concentration and the highest SR 4233 concentration. Very little metabolism occurred at 10 to 20% O2, which is in agreement with data in Chinese hamster ovary cells under aerobic conditions.

Oxygen concentration-dependent metabolism of leukotriene B4 by hepatocyte monolayers.

Leukotriene B4 was found to be metabolized by rat hepatocyte monolayers at a rate that was linear with increasing substrate concentration from 74 to 740 nM leukotriene B4. The rates of metabolism were dependent on the O2 concentration and were 315, 213, 80, and 36 pmol leukotriene B4 per min per nmol cytochrome P-450 at 20% (212 microM), 4% (42.5 microM), 2% (21.2 microM), and 1% (10.6 microM) O2, respectively. The metabolic rate was not linear with respect to O2 concentration; however, half maximal rate occurred at 4% O2, and O2 concentration found in the pericentral region of normally oxygenated liver. These results suggest that in vivo conditions of hypoxia or ischemia that lead to blood O2 concentrations less than 4% may drastically decrease hepatic clearance of leukotriene B4.

The 1,2-dichloroethylenes: their metabolism by hepatic cytochrome p-450 in vitro.

Cis- and trans-1,1-dichloroethylene bound to the active site of hepatic microsomal cytochrome P-450 with the production of a Type I difference spectrum and stimulated CO-inhibitable hepatic microsomal NADPH oxidation. Incubation of cis- and trans-1,2-dichloroethylene plus hepatic microsomes, NADPH-generating system-EDTA resulted in the production of measurable levels of 2,2-dichloroethanol and dichloroacetaldehyde but not of 2-chloroethanol, chloroacetaldehyde or chloroacetic acid and, also, resulted in decreased levels of hepatic microsomal cytochrome P-450 and heme. In addition, dichloroacetic acid was produced from trans-dichloroethylene under these experimental conditions. The omission of any component of the incubation mixture eliminated the above effects, while the inclusion of SKF-525A, metyrapone or CO:O2 (80, v/v) diminished these effects. The effects of beta-naphthoflavone and phenobarbital pretreatment on the values of Ks, delta Amax, Km and Vmax for the binding and metabolism of the 1,2-dichloroethylenes are reported. The binding and metabolism of the 1,2-dichloroethylenes and the 1,2-dichloroethylene-mediated inactivation of cytochrome P-450 were enhanced per mg of microsomal protein, but generally not per nmole of cytochrome P-450 by prior induction with beta-naphthoflavone or phenobarbital. It is concluded that multiple forms of hepatic microsomal cytochrome P-450 bind and metabolize the 1,2-dichloroethylenes. The role of cytochrome P-450 in the metabolic activation of the dichloroethylenes is considered.

Toxicity of styrene vapor in hepatocyte monolayers at low oxygen tensions.

Hepatocyte monolayer cultures were exposed to 6000 ppm styrene vapor at 20%, 2%, or 1% O2 and assayed for signs of cell damage immediately following the 2-hr exposure and then 24 hr later. Oxygen concentrations were used that were previously shown to maximize lipid peroxidation and to predispose hepatocyte monolayers to chemical injury. The use of two time points allowed assessment of acute injury as well as injury that requires several hours to manifest itself. The uptake of styrene into the buffer in the culture dishes was measured by gas chromatography and was found to be 0.49, 0.68, and 0.74 mM at 15, 60, and 120 min, respectively. However, as measured by release of aspartate aminotransferase and inclusion of trypan blue, no toxicity was evident at either time point, irrespective of the oxygen concentration. This study shows that despite the weakening of hepatocyte defense mechanisms by hypoxia, styrene is not acutely toxic to these cells. Furthermore, if any damage to DNA, RNA, or the capability for protein synthesis occurs during exposure to styrene, it is insufficient to cause lysis within 24 hr.

Ceramic primer heat-treatment effect on resin cement/Y-TZP bond strength.

The purpose of this study was to evaluate the effect of different heat-treatment strategies for a ceramic primer on the shear bond strength of a 10-methacryloyloxydecyl-dihydrogen-phosphate (MDP)-based resin cement to a yttrium-stabilized tetragonal zirconia polycrystal (Y-TZP) ceramic. Specimens measuring 4.5 × 3.5 × 4.5 mm(3) were produced from Y-TZP presintered cubes and embedded in polymethyl methacrylate (PMMA). Following finishing, the specimens were cleaned using an ultrasound device and distilled water and randomly divided into 10 experimental groups (n=14) according to the heat treatment of the ceramic primer and aging condition. The strategies used for the experimental groups were: GC (control), without primer; G20, primer application at ambient temperature (20°C); G45, primer application + heat treatment at 45°C; G79, primer application + heat treatment at 79°C; and G100, primer application + heat treatment at 100°C. The specimens from the aging groups were submitted to thermal cycling (6000 cycles, 5°C/55°C, 30 seconds per bath) after 24 hours. A cylinder of MDP-based resin cement (2.4 mm in diameter) was constructed on the ceramic surface of the specimens of each experimental group and stored for 24 hours at 37°C. The specimens were submitted to a shear bond strength test (n=14). Thermal gravimetric analysis was performed on the ceramic primer. The data obtained were statistically analyzed by two-way analysis of variance and the Tukey test (α=0.05). The experimental group G79 without aging (7.23 ± 2.87 MPa) presented a significantly higher mean than the other experimental groups without aging (GC: 2.81 ± 1.5 MPa; G20: 3.38 ± 2.21 MPa; G100: 3.96 ± 1.57 MPa), showing no difference from G45 only (G45: 6 ± 3.63 MPa). All specimens of the aging groups debonded during thermocycling and were considered to present zero bond strength for the statistical analyses. In conclusion, heat treatment of the metal/zirconia primer improved bond strength under the initial condition but did not promote stable bonding under the aging condition.

Candida parapsilosis meningitis as the first manifestation of AIDS: case report.

Candida meningitis is a rare condition that occurs more frequently in premature infants, immunocompromised patients or patients after neurosurgery. We describe a case of a previously healthy 41-year-old man with Candida parapsilosis meningitis associated with oropharyngeal candidiasis as the first manifestation of AIDS.

Mycotic aneurysm caused by Burkholderia pseudomallei: report of a Brazilian strain genetically related to Thai strains.

Melioidosis, a severe infectious disease caused by Burkholderia pseudomallei that is prevalent in Southeast Asia and Northern Australia, has been sporadically reported in Brazil since 2003. We report a case of aortic aneurysm with blood culture positive for B. pseudomallei. The phylogenetic analysis of 16S ribosomal DNA showed this isolate to be evolutionarily grouped with the MSHR346 strains from Thailand.

In vitro activities of caspofungin, amphotericin B and azoles against Coccidioides posadasii strains from Northeast, Brazil.

Coccidioidomycosis is a systemic infection caused by the soil-dwelling dimorphic fungi Coccidioides spp. The disease is endemic in semiarid Northeast Brazil, where it is caused by C. posadasii. The aim of this study was to perform antifungal susceptibility tests of clinical and environmental strains of C. posadasii from Northeast Brazil. The in vitro activities of caspofungin, amphotericin B and azoles against clinical and environment isolates of C. posadasii were determined in accordance with the NCLLS M-38P macrodilution method. The antifungal susceptibility analysis showed that all the strains of C. posadasii (n = 10) were sensitive to caspofungin (16 microg/ml < or = MIC < or = 32 microg/ml), amphotericin B (0.0625 mug/ml < or = MIC < or = 0.125 microg/ml), ketoconazole (0.039 microg/ml < or = MIC < or = 0.156 microg/ml), itraconazole (0.125 microg/ml < or = MIC < or = 0.5 microg/ml), fluconazole (3.125 microg/ml < or = MIC < or = 6.25 microg/ml), and voriconazole (0.125 microg/ml). This study is the first description of in vitro antifungal susceptibility pattern of Brazilian strains of C. posadasii.

Chlorinated ethylenes: their metabolism and effect on DNA repair in rat hepatocytes.

The major initial metabolites of the chlorinated ethylenes in hepatocyte suspensions isolated from phenobarbital treated rats were as follows (rates of metabolite production in nmol/10(6) cells/min are given in brackets): vinylidene chloride, dichloroacetic acid (0.015); cis-1,2-dichloroethylene, 2,2-dichloroethanol (0.24); trans-1,2-dichloroethylene, dichloroacetic acid (0.005); trichloroethylene, chloral hydrate (2.7); tetrachloroethylene, trichloroacetic acid (0.08). Comparison of the metabolism of the chlorinated ethylenes by isolated hepatocyte suspensions and hepatic microsomes indicates that the initial products of the three dichlorinated ethylenes from cytochrome P-450 in hepatic microsomes are rapidly and extensively metabolized in the hepatocyte, where the Phase II enzymes are present. In contrast, the initial metabolites of trichloroethylene and of tetrachloroethylene in the two systems are identical. The abilities of the chlorinated ethylenes to induce unscheduled DNA synthesis was assessed in isolated hepatocytes using a method which does not require the blocking of semi-conservative DNA synthesis. Vinylidene chloride, cis-1,2-dichloroethylene and trichloroethylene induced unscheduled DNA synthesis, while trans-1,2-dichloroethylene and tetrachloroethylene did not.

Toxicity of 1,2-dibromoethane in primary hepatocyte monolayer cultures: lack of dependence on oxygen concentration.

Hepatic 1,2-dibromoethane (DBE) metabolism proceeds via two pathways: oxidation by cytochrome P-450 and direct conjugation with the ubiquitous tripeptide glutathione (GSH) via the GSH S-transferases. The toxicity of DBE in monolayers of hepatocytes was assessed to establish whether the toxicity of this compound is increased under conditions of reductive metabolism at low oxygen concentrations. Our previous studies with t-butyl hydroperoxide and the calcium ionophore A23187 suggested that hypoxia would exacerbate toxicity that was mediated through lipid peroxidation or loss of calcium homeostasis. Monolayers of hepatocytes were exposed for 2 hr to 0, 14, 140, 1400, or 14,000 ppm of DBE in an atmosphere of either 1, 2, or 20% oxygen. Toxicity was measured by leakage of aspartate aminotransferase (AST) and trypan blue exclusion. The time course of the development of cytotoxicity was examined by assaying cell death both immediately following a 2-hr exposure and 24 hr later. The LC50 of DBE vapor was found to be approximately 14,000 ppm when assayed immediately after exposure but only 140 ppm when assayed 24 hr after exposure. The similarity of the percentages of DBE-induced cell death after incubations at 1, 2, and 20% oxygen demonstrates that the toxicity of DBE is oxygen-independent. We conclude that while DBE is highly toxic to rat hepatocytes, hypoxia does not appear to contribute to the toxicity of DBE, even under conditions of low oxygen concentrations. This result is in direct contrast to a previous report where we showed that the toxicity of halothane is potentiated under hypoxic conditions.

Interaction of hypoxia and carbon tetrachloride toxicity in hepatocyte monolayers.

The toxicity of carbon tetrachloride (CCl4) in monolayer cultures of primary hepatocytes was investigated at oxygen concentrations that prevail in the liver under conditions that range from normoxia to hypoxia: 0.5, 1, 2, and 20% O2. CCl4 was administered in the vapor phase at concentrations that produce aqueous concentrations at 37 degrees C of 0.4, 2.0, and 4.0 mM. Damage was assayed by leakage of aspartate transaminase and the inclusion of Trypan Blue immediately after the 2-hr incubation and after an additional 6-hr incubation in 20% O2. Only in the case of 0.5% O2 and 4 mM CCl4 were the monolayers damaged (18%) immediately after the 2-hr exposure; all other exposed cells were undamaged at that time point and the dose response of cell death as a function of CCl4 and oxygen concentration was not evident until the 6-hr time point. The monolayers exposed to 4 mM CCl4 and 1, 2, or 20% O2 exhibited little immediate damage but were all 100% dead 6 hr later. The monolayers exposed to 2 mM CCl4 and 0.5, 1, 2, or 20% O2 were 53, 48, 40, and 22 +/- 2% dead after 6 hr, respectively. These results suggest that effects of CCl4 exposure, for example alterations in the function or synthesis of essential proteins, require several hours to affect cell viability.

Inhibition of protein kinase C phosphorylation by mono- and divalent cations.

The effect of a matrix of concentrations of Ca2+ (0.01, 0.1, 0.5, 5 mM), Mg2+ (0.2, 0.5, 1, 2, 5, 10 mM), and Na+ (50, 100, 150 mM) on the phosphorylation of histone H-1 by protein kinase C was measured in the presence of 5 mol % diacylglycerol and Mg-ATP in both phosphatidylserine micelles and liposomes formed from a 1:4 mixture of phosphatidylserine and phosphatidylcholine. Monovalent cations (150 mM) reduced activity by 60 and 84% in the micelle and liposome assay systems, respectively. Inhibition was also observed with 5 mM Ca2+ and 10 mM Mg2+. The phosphorylating activity was compared with computer calculations of the negative electrostatic potentials (psi o) of the phospholipid membranes in the presence of the cations.

Dissociation of cytochrome P-450 inactivation and induction.

Polycyclic aromatic hydrocarbons bind to a cytosolic receptor that translocates into the nucleus and induces the synthesis of mRNA for a subset of cytochrome P-450 isozymes. The failure to detect a "phenobarbital" receptor, however, suggests that the induction of P-450b/e, the isozymes induced by phenobarbital, is mediated by a less direct mechanism. One attractive alternative is that a receptor exists for an endogenous substance that is specifically turned over by the trace amounts of P-450b/e present in uninduced liver. Changes in the concentration of this substance caused by agents that inhibit P-450b/e would then trigger the induction response. We report here that suppressing the catalytic activities of P-450b/e for prolonged periods with 1-aminobenzotriazole, a mechanism-based irreversible inhibitor, neither induces P-450b/e nor interferes with induction of these isozymes at the level of transcription by phenobarbital. The results argue against mechanisms for the induction of P-450b/e keyed to the effective catalytic availability of these isozymes.

Comparative toxicity of halothane, isoflurane, hypoxia, and phenobarbital induction in monolayer cultures of rat hepatocytes.

Hypoxia, phenobarbital induction, and halothane anesthesia have been implicated in the pathogenesis of hepatotoxicity in the rat model. However, a controversy exists over the role of halothane in liver injury; does it act by reducing hepatic blood flow, thereby inducing hypoxia, or do its metabolites initiate the injury? These variables are difficult to separate during in vivo halothane exposure. In the present experiments, effects of halothane on hepatic perfusion were eliminated by exposing confluent monolayers of hepatocytes isolated from Fisher 344 rats livers, both with and without phenobarbital pretreatment, to 1.5% halothane or 2.0% isoflurane in 1%, 2%, or 4% (control) oxygen. Isoflurane exposure was included for a control of anesthetic effects on hepatocytes, because it is known to be metabolized minimally and probably is not associated with hepatic dysfunction. Oxygen levels were chosen to approximate those that may occur in the liver in vivo. Cell death was assayed via aspartate aminotransferase (AST) release, both immediately following a 2-h oxygen +/- anesthetic exposure and 6 h post-exposure. Per cent cell death data were analyzed using multiple regression techniques. Results obtained immediately, and 6 h after, exposure demonstrate that low oxygen levels, halothane, and phenobarbital were each highly significant factors (P less than .001) in relation to cell death, in agreement with the halothane-phenobarbital-hypoxia rat model. A toxic effect of isoflurane was not observed under identical experimental conditions. The results of the study clearly indicate that the origin of cell death in hepatocyte monolayers is multi-factorial; hypoxia, phenobarbital induction, and halothane exposure each contribute to the hepatocyte damage observed in our in vitro model.

Toxicity of calcium ionophore A23187 in monolayers of hypoxic hepatocytes.

Increased cytoplasmic calcium has been implicated in hepatic necrosis induced by cytochrome P-450-mediated halocarbon metabolism; hepatic necrosis is exacerbated hypoxia. It is known that addition of the calcium ionophore A23187 to a cell can mimic the metabolic causes of increased cytoplasmic calcium. In the present experiments, confluent monolayers of rat hepatocytes were exposed to A23187 (0.5-50 microM) under conditions of normoxia and 18-fold during a 4-h incubation at 0.5% (5 microM) O2 compared with 20% (212 microM) O2. The ED50 values of A23187 were 0.92, 1.84, 10.98, and 16.81 microM at 0.5, 1, 2, and 20% O2, respectively. The results demonstrate that small reductions in the oxygen concentrations normally encountered by pericentral hepatocytes (i.e., below 28 microM) may severely decrease the ability of these already compromised hepatocytes to manage perturbations in calcium homeostasis.