Use of a Latent Topic Model for Characteristic Extraction from Health Checkup Questionnaire Data.

OBJECTIVES: When patients complete questionnaires during health checkups, many of their responses are subjective, making topic extraction difficult. Therefore, the purpose of this study was to develop a model capable of extracting appropriate topics from subjective data in questionnaires conducted during health checkups. METHODS: We employed a latent topic model to group the lifestyle habits of the study participants and represented their responses to items on health checkup questionnaires as a probability model. For the probability model, we used latent Dirichlet allocation to extract 30 topics from the questionnaires. According to the model parameters, a total of 4381 study participants were then divided into groups based on these topics. Results from laboratory tests, including blood glucose level, triglycerides, and estimated glomerular filtration rate, were compared between each group, and these results were then compared with those obtained by hierarchical clustering. RESULTS: If a significant (p < 0.05) difference was observed in any of the laboratory measurements between groups, it was considered to indicate a questionnaire response pattern corresponding to the value of the test result. A comparison between the latent topic model and hierarchical clustering grouping revealed that, in the latent topic model method, a small group of participants who reported having subjective signs of urinary disorder were allocated to a single group. CONCLUSIONS: The latent topic model is useful for extracting characteristics from a small number of groups from questionnaires with a large number of items. These results show that, in addition to chief complaints and history of past illness, questionnaire data obtained during medical checkups can serve as useful judgment criteria for assessing the conditions of patients.

Prediction model for glucose metabolism based on lipid metabolism.

OBJECTIVES: We developed a robust, long-term clinical prediction model to predict conditions leading to early diabetes using laboratory values other than blood glucose and insulin levels. Our model protects against missing data and noise that occur during long-term analysis. METHODS: RESULTS of a 75-g oral glucose tolerance test (OGTT) were divided into three groups: diabetes, impaired glucose tolerance (IGT), and normal (n = 114, 235, and 325, respectively). For glucose metabolic and lipid metabolic parameters, near 30-day mean values and 10-year integrated values were compared. The relation between high-density lipoprotein cholesterol (HDL-C) and variations in HbA1c was analyzed in 158 patients. We also constructed a state space model consisting of an observation model (HDL-C and HbA1c) and an internal model (disorders of lipid metabolism and glucose metabolism) and applied this model to 116 cases. RESULTS: The root mean square error between the observed HbA1c and predicted HbA1c was 0.25. CONCLUSIONS: In the observation model, HDL-C levels were useful for prediction of increases in HbA1c. Even with numerous missing values over time, as occurs in clinical practice, clinically valid predictions can be made using this state space model.

Hepatocyte destruction with enhanced collagen synthesis: characteristic feature of chronic hepatitis C patients on haemodialysis.

Hepatitis C virus (HCV) infection is frequent among patients with end-stage renal disease on haemodialysis and is considered to be an independent risk factor for mortality in this setting. However, only a few of these patients are treated with anti-hepatitis virus treatment before the development of end-stage renal disease. Recent guidelines recommend identification of patients with good prognoses who are in need of interferon treatment, but we know little of patients who must be treated urgently. Ninety-eight patients on haemodialysis (48 anti-HCV-positive and 50 anti-HCV-negative patients) were enrolled in this study; HCV RNA was detected in 43 anti-HCV-positive patients. Univariate analysis and multivariate regression analysis were applied to identify variables independently associated with persistent HCV infection. Seven variables were proven to be associated with persistent HCV infection. Among them, type IV collagen 7S and N-terminal propeptide of type III procollagen (P-III-P) were defined as independent variables useful in distinguishing HCV RNA-positive patients from HCV RNA-negative patients with 0.91 sensitivity, 0.91 specificity, 0.89 positive predictive value and 0.93 negative predictive value. Our observations suggest that hepatocyte destruction with enhanced liver fibrosis is a characteristic clinical feature of persistent HCV infection. Type IV collagen 7S of ≥ 5 ng/mL and/or P-III-P of ≥ 5 U/mL would be useful markers to identify patients in need of interferon treatment, which supports the idea of the Kidney Disease: Improving Global Outcomes guidelines that a good prognosis in patients with HCV infection on haemodialysis should prompt consideration for IFN treatment when applicable.

Fifty-two week chronic toxicity of enzymatically decomposed rutin in Wistar rats.

The chronic toxicity of enzymatically decomposed rutin, which consists mainly of isoquercitrin, was evaluated in male and female Wistar rats with dietary administration at concentrations of 0%, 0.04%, 0.2%, 1% and 5% for 52 weeks. No toxicological findings were found in the mortality, body weights, food consumption, hematology, clinical biochemistry or organ weights in either sex. Obvious clinical signs were chromaturia that could be attributed to the color of test substance in the 5% groups of both sexes. Coloration of the urine collected over 24h in the 1% and 5% groups of both sexes was noted. Increased daily urinary calcium excretion was observed in the 5% groups of both sexes and an increase in urinary calcium concentration was observed in the male 5% group. On histopathological examination, incidences of mineralization, inflammatory cell debris, inflammatory cell infiltration and/or transitional cell hyperplasia in the renal pelvis were increased in the 5% male group, whereas treated females showed no apparent difference in these incidences. Based on the above findings, the no observed adverse effect level (NOAEL) was estimated to be 1% in both sexes (542.4 mg/kg body weight/day for males and 674.0mg/kg body--weight/day for females).

Full implementation of a computerized nursing records system at Kochi Medical School Hospital in Japan.

A project to fully implement a novel computerized nursing records system resulted in the standardization of nursing records, reduced the administrative workload for staff, enabled medical staff to know a patient's status at any given time, and improved the quality of nursing care provided to patients. The development process of the computerized nursing records system involved three main steps: 1) the establishment of a new nursing assessment form and introduction of nursing diagnosis into routine work, 2) computerized system design and construction, and 3) the usability check of the computerized nursing records system in a clinical setting for 1 year. The successful development of the computerized nursing records system was based on the following points: 1) the assessment form for nursing diagnosis was improved and the nursing diagnosis was introduced before the computerized system was designed and constructed; 2) full, rather than partial, implementation of the computerized system occurred; 3) existing knowledge of nursing assessments and standard care planning were fully used; 4) registered data were optimally reused upon summarization and readmission to reduce the nurses' workload; and 5) portable computers were introduced to enable simple and quick recording of bedside findings. The routine use of the computerized nursing records system was started in April 2000. More comprehensive investigations during the next 2 to 3 years are necessary to determine how the contents of nursing records can be improved and how much the computerized nursing records system affects the quality of nursing.

A new theory of phylogeny inference through construction of multidimensional vector space.

Here, a new theory of molecular phylogeny is developed in a multidimensional vector space (MVS). The molecular evolution is represented as a successive splitting of branch vectors in the MVS. The end points of these vectors are the extant species and indicate the specific directions reflected by their individual histories of evolution in the past. This representation makes it possible to infer the phylogeny (evolutionary histories) from the spatial positions of the end points. Search vectors are introduced to draw out the groups of species distributed around them. These groups are classified according to the nearby order of branches with them. A law of physics is applied to determine the species positions in the MVS. The species are regarded as the particles moving in time according to the equation of motion, finally falling into the lowest-energy state in spite of their randomly distributed initial condition. This falling into the ground state results in the construction of an MVS in which the relative distances between two particles are equal to the substitution distances. The species positions are obtained prior to the phylogeny inference. Therefore, as the number of species increases, the species vectors can be more specific in an MVS of a larger size, such that the vector analysis gives a more stable and reliable topology. The efficacy of the present method was examined by using computer simulations of molecular evolution in which all the branch- and end-point sequences of the trees are known in advance. In the phylogeny inference from the end points with 100 multiple data sets, the present method consistently reconstructed the correct topologies, in contrast to standard methods. In applications to 185 vertebrates in the alpha-hemoglobin, the vector analysis drew out the two lineage groups of birds and mammals. A core member of the mammalian radiation appeared at the base of the mammalian lineage. Squamates were isolated from the bird lineage to compose the outgroup, while the other living reptilians were directly coupled with birds without forming any sister groups. This result is in contrast to the morphological phylogeny and is also different from those of recent molecular analyses.

New approach to the medical information system for quality management in patient care: development of Problem Mapping System.

A new type of medical information system named Problem Mapping System (P-Map) has been developed, which aids physicians with solving patients' problems. With this system, physicians can define the problems of in-patients, monitor their progress clearly, and share information efficiently. In P-map, a list of problems, such as disease names, can be set for each inpatient easily. The progress of each problem is clearly shown using progress lines on a time axis. Physicians can save the Subjective Objective Assessment Plan (SOAP) notes which are linked to each problem. At the final stage of patient care, a discharge summary can be made easily. With the aid of this system, the quality of patient care is improved due to the following: (1) physicians can make the best decision; (2) medical staff in the same team can provide the best medical treatment; (3) evaluation of each medical treatment is easy; (4) saved data can be used effectively for education and research; (5) the system can improve cooperation with other medical institutes by providing discharge summary information which can be distributed using e-mail; and (6) the system can improve patients' understanding for the purpose of informed consent by providing clear and well organized information to patients.

Problems regarding the integration of medical images into the total hospital information system.

Since 1989, the integration of medical images into the total hospital information system (HIS) has been investigated and developed at Kochi Medical School. The basic concept of the integration is that, in the same way they can view text based data, doctors can retrieve and view images using the PC terminals of the total HIS. The possibility of utilizing the PC terminals of the total HIS as image viewing stations was investigated. A test run was performed in the period from October 1995 to July 1997. The test run revealed that fast image access is crucial in order for the system to be useful for doctors. After making various improvements, the final system became well used in the clinical practice. However, in order to progress to the film-less stage, the final system still has three problems that must be solved: quality of the image display, operation of multi-exams, and quality assurance of the digital image.

Fast image accessing based on an analysis of the picture archiving and communications system in Kochi Medical School.

A large-scale picture archiving and communication system was developed by Kochi Medical School. Using JPEG to achieve 90% compression, this system was capable of storing several years worth of CT and MR examination records and eliminated the bottleneck in conventional systems caused by slow data access in the filing system. Under the new system, access of image files from client PCs was checked for bottlenecks. The image file retrieval from the image database on the server was found to require the longest time. Two methods were considered to eliminate this bottleneck. One was a multiimages filing into a file. Another was an image transmission with multiparallel transmission sessions, which is proposed in this article. By adopting the latter method, the system can transmit 50 CT images from the image database server to a client PC in about 10 s.

Waveform changes due to conduction block and their underlying mechanism in spinal somatosensory evoked potential: a computer simulation. Technical note.

Based on a square-wave solid-angle analysis, a simplified mathematical model was produced for computing a sequence of potential change in a volume conductor generated by an impulse traveling along a nerve fiber. A conduction block was simulated as a phenomenon in which a depolarization wavefront stops traveling when it reaches a certain point, although the following repolarization wavefront continues to travel until it reaches the same point. The spinal somatosensory evoked potential (SSEP) was produced as an algebraic sum of simulated nerve fiber action potentials (NFAPs). With a conduction block, an NFAP that was normally triphasic showed a positive-negative diphasic wave with reduced negativity at the point of the block, diphasic waves with enhanced negativity at points immediately preceding the block, and initial-positive waves alone or abolition of any wave at points beyond the block. The absence of their terminal-positive phases paradoxically enhanced the negative peak of the spinal SSEPs in a partial block that involved only the constituent fastest fibers, because phase cancellation of the phases between the terminal-positive phases of the fastest fibers and the negative phases of the slower fibers, which normally happens, failed to occur. At the points immediately preceding the block, the identical mechanism sustained the spinal SSEP enhancement even when every fiber was included in the block. The computer model predicted that localization of the precise site of conduction block can be achieved by demonstrating an abrupt reduction in the amplitude of the spinal SSEP, which is accompanied by an increased negative wave caudally and an enhanced monophasic positive wave rostrally.

Increase or decrease of HDL-cholesterol concentrations during pravastatin treatment depending on the pre-treatment HDL cholesterol levels.

OBJECTIVE: The effect of pravastatin was evaluated using patient data accumulated in the data base of a hospital information system (HIS). METHODS: We selected 130 patients treated with pravastatin 10 mg per day, for a minimum period of 4 weeks. RESULTS: In the t test analysis, the reduction rates of total cholesterol (TC) and low-density lipoprotein (LDL) levels for pravastatin administration were 18%, and 27%, respectively. These values were similar to previous reports. The high-density lipoprotein (HDL) level, however, did not change significantly, although previous reports have shown an elevation of HDL levels. In an attempt to explain the origin of this difference, we studied the pretreatment value dependence of the cholesterol change using regression analysis. We found that pravastatin raised the HDL level in those cases where pretreatment values were lower than 58 mg.dl-1 and reduced it for higher values. We also showed that the reductions of TC, LDL and triglyceride (TG) levels correlated positively with their pretreatment values.

Longitudinal characteristic curve of liver disease.

A longitudinal characteristic curve of chronic liver disease (LCC-LD) is derived for the first time by a new method of time series data analysis, where a hospital information system is utilized as a clinical research application of the database. It describes a typical pattern of development of disease from the beginning of chronic hepatitis to the final stage of cirrhosis. The LCC-LD is obtained by effectively using patient data with various stages of developments of liver disease and the present method is applicable to derive the LCC of other diseases. The obtained LCC-LD may be useful for a clinical decision making support such as the prospective assessment (for example, the onset time of cirrhosis) of liver disease in individual patients, an evaluation of drug effect, etc.

Dynamical monitoring network system for perinatal care.

A new type of monitoring network system for perinatal care is proposed and has been developed. The patient monitoring system and data analyzing system are connected by a local area network (LAN). The doctor can retrieve past sampled data and results of data analysis, and make a detailed analysis at the same time that the patient is being monitored. The mainframe of the hospital information system (HIS) is connected with a data server in the perinatal care area via LAN. The database in perinatal care is supplied to the HIS and the doctor through this monitoring network system. If data sampled at maternity clinics or hospitals are once transmitted to the data server on LAN via the public telephone circuit, these data are available for the specialists on LAN. This function is utilized for supporting the obstetricians.

Development of a diagnostic and therapeutic simulation system based on patient data and specialist's knowledge. I. Simulation of diagnostic process.

A new simulation system of diagnostic and therapeutic processes is developed. The aim is to train medical students for the practical use of their knowledge, utilizing patient data in a total hospital information system. The knowledge in the system is presented by the specialists for every case. In medical school there are many specialists in various fields. With their cooperation the system can grow up to a comprehensive CAI system for clinical education. The system is designed to work on the mainframe for easiness of development, maintenance and extensions of the system. The present framework has been applied to the simulation of diagnostic process. The usefulness of the present system has been confirmed by specialists and students.