A combined Fuzzy and Naive Bayesian strategy can be used to assign event codes to injury narratives.

BACKGROUND: Bayesian methods show promise for classifying injury narratives from large administrative datasets into cause groups. This study examined a combined approach where two Bayesian models (Fuzzy and Naïve) were used to either classify a narrative or select it for manual review. METHODS: Injury narratives were extracted from claims filed with a worker's compensation insurance provider between January 2002 and December 2004. Narratives were separated into a training set (n=11,000) and prediction set (n=3,000). Expert coders assigned two-digit Bureau of Labor Statistics Occupational Injury and Illness Classification event codes to each narrative. Fuzzy and Naïve Bayesian models were developed using manually classified cases in the training set. Two semi-automatic machine coding strategies were evaluated. The first strategy assigned cases for manual review if the Fuzzy and Naïve models disagreed on the classification. The second strategy selected additional cases for manual review from the Agree dataset using prediction strength to reach a level of 50% computer coding and 50% manual coding. RESULTS: When agreement alone was used as the filtering strategy, the majority were coded by the computer (n=1,928, 64%) leaving 36% for manual review. The overall combined (human plus computer) sensitivity was 0.90 and positive predictive value (PPV) was >0.90 for 11 of 18 2-digit event categories. Implementing the 2nd strategy improved results with an overall sensitivity of 0.95 and PPV >0.90 for 17 of 18 categories. CONCLUSIONS: A combined Naïve-Fuzzy Bayesian approach can classify some narratives with high accuracy and identify others most beneficial for manual review, reducing the burden on human coders.

Bayesian methods: a useful tool for classifying injury narratives into cause groups.

To compare two Bayesian methods (Fuzzy and Naïve) for classifying injury narratives in large administrative databases into event cause groups, a dataset of 14 000 narratives was randomly extracted from claims filed with a worker's compensation insurance provider. Two expert coders assigned one-digit and two-digit Bureau of Labor Statistics (BLS) Occupational Injury and Illness Classification event codes to each narrative. The narratives were separated into a training set of 11 000 cases and a prediction set of 3000 cases. The training set was used to develop two Bayesian classifiers that assigned BLS codes to narratives. Each model was then evaluated for the prediction set. Both models performed well and tended to predict one-digit BLS codes more accurately than two-digit codes. The overall sensitivity of the Fuzzy method was, respectively, 78% and 64% for one-digit and two-digit codes, specificity was 93% and 95%, and positive predictive value (PPV) was 78% and 65%. The Naïve method showed similar accuracy: a sensitivity of 80% and 70%, specificity of 96% and 97%, and PPV of 80% and 70%. For large administrative databases, Bayesian methods show significant promise as a means of classifying injury narratives into cause groups. Overall, Naïve Bayes provided slightly more accurate predictions than Fuzzy Bayes.

The effect of diphenylhydantoin on thyroxine metabolism in man.

The effect of 5,5'-diphenylhydantoin on thyroxine metabolism was examined in five normal volunteers. Intravenous injection of radiothyroxine was followed by a 10-12 day control and subsequent 9-14 day treatment periods. During oral administration of diphenylhydantoin, plasma thyroxine concentration decreased to about 80% of its pretreatment level and the plasma radiothyroxine disappearance rate increased a maximum of 20% over control estimates. These changes were a result of increases in both urinary and fecal excretion of radioisotope.A minimum plasma thyroxine was apparent after 10-12 days of diphenylhydantoin administration. In two of the subjects, treatment was sufficiently prolonged to achieve this new steady state. In these subjects, the decrease in total body thyroxine was balanced by the increase in the fractional turnover rate. As a result, absolute thyroxine degradation during diphenylhydantoin administration was unchanged from the pretreatment values. Plasma ultrafiltration was used to estimate the free thyroxine fraction at regular intervals during the control and treatment periods. During diphenylhydantoin treatment, there was little or no change in this fraction and therefore, absolute free thyroxine decreased. Thyroxine-binding globulin and thyroxine-binding prealbumin capacities remained constant. These results indicate that thyroxine degradation can proceed at a normal rate in subjects receiving diphenylhydantoin despite decreases in plasma free thyroxine concentration. If free thyroxine is the only portion of the hormone available for cellular utilization, then free thyroxine clearance must be increased in these subjects. This increase in clearance could represent either a direct stimulation of peripheral thyroxine metabolism by diphenylhydantoin, or it could reflect the response of intrinsic regulatory systems to a diphenylhydantoin-mediated displacement of thyroxine from thyroxine-binding globulin. Whatever the mechanism for this effect, a decreased free thyroxine value in patients receiving diphenylhydantoin may not imply hypothyroidism.

Left ventricular function in tricuspid atresia: a radionuclide study.

Left ventricular function was evaluated by radionuclide angiocardiography in patients with tricuspid atresia before and after the Fontan operation. Ejection fraction was 0.54 +/- 0.13 in 29 patients without the Fontan procedure and 0.56 +/- 0.14 in 14 postoperative patients. Group means were not statistically different, but both were less than normal values for the institution (p less than 0.001). Function did not correlate significantly with age or aortic oxygen saturation. Serial studies demonstrated little change in eight of nine patients examined over 1 to 3 years preoperatively. From the preoperative group of nine patients with abnormal ejection fraction, three underwent surgery, with two survivors. Although 5 of 14 patients had abnormal systolic function after the Fontan operation, only 1 was symptomatic. Depressed response to isometric exercise was found in two patients with a normal ejection fraction at rest. Group systolic function after Fontan surgery did not correlate significantly with age at surgery but all patients were operated on at more than 5 years of age. Of seven patients studied both preoperatively and postoperatively two had an increase and three a decrease in ejection fraction. The trend in these seven studied serially after surgery was toward early improvement in ejection fraction. This study shows variability in left ventricular performance in patients with tricuspid atresia before and after surgery. Individuals undergoing the Fontan procedure after 5 years of age do not necessarily have an increase in ejection fraction over preoperative values, and may even have a decrease. Serial improvement over early postoperative results may be expected, without significant deterioration, during the next 1 to 3 years.

Regional cardiac sympathetic denervation in patients with ventricular tachycardia in the absence of coronary artery disease.

OBJECTIVES: The aim of this study was to determine whether patients with ventricular arrhythmias in the absence of coronary artery disease also have abnormalities in sympathetic innervation. BACKGROUND: We have previously shown by cardiac sympathetic scintigraphy using iodine-123-metaiodobenzylguanidine (I-123-MIBG) that patients with ventricular tachycardia after myocardial infarction have regional cardiac sympathetic denervation. It is not known whether patients with ventricular tachycardia in the absence of coronary artery disease also have regional cardiac sympathetic denervation. METHODS: We performed cardiac I-123-MIBG and thallium-201 single-photon emission computed tomographic (SPECT) scans at rest in 18 patients (mean age 47 +/- 18 years) with cardiomyopathy (n = 6), left ventricular hypertrophy (n = 1), valvular disease (n = 2) or a structurally normal heart (n = 9) who presented with monomorphic (n = 15) or polymorphic (n = 3) ventricular tachycardia. These scans were compared with scans in 12 control patients without ventricular tachycardia (mean age 30 +/- 17 years) who had cardiomyopathy (n = 3) or a structurally normal heart (n = 9). Cardiac sympathetic denervation was defined as myocardial areas having thallium uptake with reduced or absent I-123-MIBG uptake. RESULTS: Twelve (67%) of 18 patients with ventricular tachycardia had regional cardiac sympathetic denervation compared with 1 (8%) of 12 patients who did not have ventricular tachycardia (p = 0.002). In the nine patients with a structurally normal heart and ventricular tachycardia, five (55%) patients had regional cardiac sympathetic denervation compared with zero of nine control patients with a structurally normal heart (p = 0.029). Five patients underwent right ventricular radiofrequency ablation for ventricular tachycardia, and sympathetic denervation was adjacent to the ablation site in one of these patients. CONCLUSIONS: Patients with ventricular tachycardia in the absence of coronary artery disease have abnormal cardiac sympathetic innervation detectable by cardiac sympathetic scintigraphy. The role of regional cardiac sympathetic denervation in arrhythmogenesis remains to be determined.

Regional sympathetic denervation after myocardial infarction in humans detected noninvasively using I-123-metaiodobenzylguanidine.

Transmural myocardial infarction in dogs produces denervation of sympathetic nerves in viable myocardium apical to the infarct that may be arrhythmogenic. It is unknown whether sympathetic denervation occurs in humans. The purpose of this study was to use iodine-123-metaiodobenzylguanidine (MIBG), a radiolabeled guanethidine analog that is actively taken up by sympathetic nerve terminals, to image noninvasively the cardiac sympathetic nerves in patients with and without ventricular arrhythmias after myocardial infarction. Results showed that 10 of 12 patients with spontaneous ventricular tachyarrhythmias after myocardial infarction exhibited regions of thallium-201 uptake indicating viable perfused myocardium, with no MIBG uptake. Such a finding is consistent with sympathetic denervation. One patient had frequent episodes of nonsustained ventricular tachycardia induced at exercise testing that was eliminated by beta-adrenoceptor blockade. Eleven of the 12 patients had ventricular tachycardia induced at electrophysiologic study and metoprolol never prevented induction. Sympathetic denervation was also detected in two of seven postinfarction patients without ventricular arrhythmias. Normal control subjects had no regions lacking MIBG uptake. This study provides evidence that regional sympathetic denervation occurs in humans after myocardial infarction and can be detected noninvasively by comparing MIBG and thallium-201 images. Although the presence of sympathetic denervation may be related to the onset of spontaneous ventricular tachyarrhythmias in some patients, it does not appear to be related to sustained ventricular tachycardia induced at electrophysiologic study.

Ictal brain imaging in presurgical evaluation of patients with medically intractable complex partial seizures.

At the Indiana University Medical Center, 99 patients with medically intractable complex partial seizures (MI-CPS) had presurgical evaluation with subsequent anterior temporal lobectomy. The majority of the patients had single photon emission tomography (SPECT) performed interictally as well as during an actual epileptic seizure (ictal scan). Decreased regional cerebral perfusion (rCP) was seen in 54/94 (57%) of the interictal scans corresponding to the eventual site of the surgery. However, ictal scans provided a higher yield; increased rCP in the temporal lobe during an actual seizure was observed in 60/82 (73%) concordant to the side of surgery. SPECT is a useful, noninvasive method of localizing the epileptiform focus in patients with MI-CPS considered for resective surgery. Both interictal and ictal SPECT need to be performed; combined interictal hypoperfusion and ictal hyperperfusion in the same focal area are unique to epileptogenic lesions. Ictal SPECT studies can be performed in the majority of patients during the period of continuous video/EEG monitoring with only a little additional effort. Combining the results of functional brain imaging (interictal and ictal SPECT, PET) with clinical semiology of seizures, surface and sphenoidal EEG, magnetic resonance imaging and other non-invasive tests, anterior temporal lobectomy can be recommended in approximately two-thirds of the patients without resorting to potentially dangerous intracranial EEG monitoring.

HIPDM-SPECT in patients with medically intractable complex partial seizures. Ictal study.

Both interictal and ictal N,N,N'-trimethyl-N'-(2-hydroxy-3-methyl-5-iodobenzyl)-1,3,propanediamine -single photon emission computed tomography (HIPDM-SPECT) were performed in 16 patients with medically intractable complex partial seizures. Ictal HIPDM-SPECT localized epileptic foci in 13 of 14 patients with unilateral temporal focus and provided confirmative evidence of epileptic focus in 11 patients by demonstrating maximally increased regional cerebral perfusion (rCP) in epileptic foci that had shown decreased rCP in a previous interictal study. Ictal HIPDM-SPECT in two patients with bitemporal foci showed more complicated patterns consisting of slightly increased rCP in bilateral multifocal regions. Ictal HIPDM-SPECT was particularly useful for investigating epileptic foci, and correlation with simultaneously recorded ictal electroencephalograms provided further insight for localizing epileptic foci.

Single photon emission computed tomography (SPECT) brain imaging using N,N,N'-trimethyl-N'-(2 hydroxy-3-methyl-5-123I-iodobenzyl)-1,3-propanediamine 2 HCl (HIPDM): intractable complex partial seizures.

HIPDM-SPECT brain imaging was performed in four patients with intractable complex partial seizures (CPS). Three patients had an epileptogenic focus in one temporal lobe and underwent anterior temporal lobectomy. Interictal HIPDM-SPECT demonstrated decreased regional cerebral perfusion (rCP) in the epileptogenic area in only one patient, but ictal studies showed increased rCP in the epileptic foci of all three patients. In the fourth patient, interictal HIPDM-SPECT showed increased rCP in the area of epileptogenic focus; when antiepileptic medication was taken, rCP decreased. HIPDM-SPECT brain imaging is useful for localizing epileptogenic foci in CPS.

Culture, variation, and levels of analysis in folk psychologies: comment on Lillard (1998)

Not only do folk psychologies differ for adults in different cultures, but naive psychological conceptions begin early in life and develop. Understanding cultural variation requires understanding these beginnings and developments as well as considering naive psychological conceptions at several different levels of analysis.

Stoichiometric Tc-99m RBC labeling using stable kit solutions of stannous chloride and EDTA: concise communication.

An in vitro Tc-99m labeling method is described, which utilizes stable stock solutions of stannous chloride and disodium edetate (EDTA). The kit procedure requires as little as 1 ml of patient blood, can be performed in only 15 min, and gives labeling yields in excess of 98%. By using EDTA, the binding capacity of RBCs for technetium is sufficient to produce Tc-99m RBC doses with specific concentration greater than 50 mCi/ml for first-pass cardiac studies. Scintigrams reveal a slight amount of bladder activity and splenic uptake, but at no time has the thyroid, stomach, or normal bowel been visualized. The predominant 20-hr blood-clearance half-time results in excellent image quality for over 24 hr--an essential property for following intermittent GI bleeding or for performing repeat cardiac function studies over a several-hour time interval.

Liver and kidney imaging with Ga-68-labeled dihydroxyanthraquinones.

This paper describes the preparation of alizarin (1,2-dihydroxyanthraquinone) and alizarin red S (sodium 1,2-dihydroxyanthraquinone-3-sulfonate) labeled with Ga-68, which is obtained from a new high-yield Ge-68 leads to Ga-68 generator. The uptake of Ga-68 alizarin by liver and spleen RES was studied in rats, dogs, and human, and amounted to 80-85% of the administered dose within 5 min after i.v. injection. Gallium-68 alizarin red S was preferentially accumulated in the renal parenchyma to an extent of 70% within 2 hr after i.v. administration. Both substances combine simple and fast preparation with the potential advantages of positron scintigraphy. Complete labeling of 1 mCi Ga-68 was achieved by 100 micrograms of each compound, amounts that are without any known measurable harm to humans (LD50 alizarin red S for i.v. injected mice = 70 mg/kg (8); LD50 alizarin for i.p. injected mice > 120 mg/kg (18)).

An automated method for the alignment of image pairs.

The computer comparison of two images of the same organ requires proper alignment of the images before further computer processing. This alignment can be achieved by (a) fixing patient position during the study, (b) alignment methods using analytical transformations, or (c) operator interaction. We propose an automated method based upon the cross-correlation between projections of the images. With fast Fourier transforms, the algorithm becomes computationally cheap.

Scintigraphic and X-ray arthrographic diagnosis of femoral prosthesis loosening: concise communication.

In order to determine loosening of femoral components of conventional total hip arthroplasties, a comparison of radionuclide (RA) and radiographic arthrograms (XA) was performed. The hips of 29 patients were studied with radiographic contrast and intraarticular Tc-99m sulfur colloid. The XAs were positive for loosening in 10/29 studies, whereas the RAs were positive in 19/29 studies. In five of ten in which both studies were positive, the RA showed femoral prosthesis loosening more clearly than the XA. By surgical follow-up, there were four false-negative XAs . Our data suggest that the RA is valuable in determining femoral component loosening.

Evaluation of bone-marrow scanning with technetium-99m sulfur colloid in pediatric oncology.

Eighty-six technetium-99m sulfur colloid (Tc-SC) bone-marrow scans in 56 pediatric oncology patients were reviewed. The distribution of the sulfur colloid was similar to that in adult bone marrow in normal children older than 10 yr, and involved progressively more marrow of the extremities in normal children under 10 years of age. After irradiation or chemotherapy there was an extension of the Tc-SC to peripheral marrow sites. There was also diminished uptake of the tracer in sites corresponding to irradiated areas. In most patients there was recovery of these defects by 6 mo after completion of therapy. Tumor replacement of the marrow was reflected in the scans, and the extent of the scan defect paralleled the course of the disease. In four patients, despite normal bone scans and radiographs, marrow-scan abnormalities due to tumor replacement were present and confirmed by needle aspiration and/or biopsy. In two other patients, the marrow-scan abnormality preceded radiographic and histologic evidence of tumor metastasis. Two patients who responded clinically showed persistent defects; biopsy in one revealed fibrosis. Technetium-99m sulfur colloid bone-marrow scanning appears to be a sensitive monitor of marrow alteration caused by metastases, irradiation damage, or tissue fibrosis in children receiving treatment for cancer.

Localization of technetium-99m pertechnetate in peripheral nerve tumors.

Technetium-99m pertechnetate scintigraphy in three patients with pathologically proven peripheral nerve tumors (six in total) demonstrated its ability to assess accurately the size, location, and the extent of all lesions. Pertechnetate localized only in areas of abnormal nerve involvement and all lesions were better seen in delayed images than the earlier ones. Pertechnetate imaging appears to be a promising method of noninvasive evaluation of clinically evident and occult tumors of peripheral nerve origin.

A quantitative evaluation of rheumatoid arthritic activity with Tc-99m HEDP.

In an attempt to develop a quantitative method of evaluating rheumatoid arthritic activity, Tc-99m HEDP joint uptake values and joint-to-bone ratios were studied in ten adult rheumatoid arthritic patients and 17 nonarthritic patients. A joint-to-bone activity ratio of 1.8 at the fourth hour after injection (RA Index) discriminated clinically active rheumatoid arthritic joints from control joints with 95% accuracy. Serial studies on five patients during drug trial demonstrated a positive correlation between RA Index and the clinical manifestations of rheumatoid arthritic activity. The RA Index may be a useful quantitative parameter for evaluation of rheumatoid arthritic activity following therapy.

Single-photon emission computed tomography (SPECT) for assessment of hepatic lesions.

Single-photon emission computed tomography (SPECT) and conventional scintigraphy were compared in 130 patients examined to assess hepatic involvement in malignant disease. Transmission computed tomography (TCT) served as the reference method against which SPECT and conventional scintigraphy were compared. The sensitivity of SPECT was calculated for lesions grouped according to diameter as well as location. The Bayesian theorem was used to assess the reliability of both SPECT and conventional scintigraphy. SPECT identified only 52% of lesions with a diameter of 1.5-2.0 cm. It was also shown that the sensitivity of SPECT was lowest for small lesions in the middle third of the liver. A comparison of the final diagnosis demonstrated that SPECT had greater sensitivity, specificity, and accuracy than conventional scans, and is superior at low disease prevalence. At high disease prevalence, SPECT has a lower rate of false negatives. SPECT appears to be the superior imaging modality for evaluation of the liver in malignant disease.

Combined bone scintigraphy and indium-111 leukocyte scans in neuropathic foot disease.

It is difficult to diagnose osteomyelitis in the presence of neurotrophic osteoarthropathy. We performed combined [99mTc]MDP bone scans and indium-111 (111In) leukocyte studies on 35 patients who had radiographic evidence of neuropathic foot disease and clinically suspected osteomyelitis. The [111In]leukocyte study determined if there was an infection and the bone scan provided the anatomic landmarks so that the infection could be localized to the bone or the adjacent soft tissue. Seventeen patients had osteomyelitis and all showed increased [111In]leukocyte activity localized to the bone, giving a sensitivity of 100%. Among the 18 patients without osteomyelitis, eight had no accumulation of [111In]leukocytes, seven had the [111In]leukocyte activity correctly localized to the soft tissue, two had [111In]leukocyte activity mistakenly attributed to the bone, and one had [111In]leukocyte accumulation in a proven neuroma which was mistakenly attributed to bone. These three false-positive results for osteomyelitis reduced the specificity to 83%. Considering only the 27 patients with a positive [111In]leukocyte study, the combined bone scan and [111In]leukocyte study correctly localized the infection to the soft tissues or bone in 89%. Uninfected neurotrophic osteoarthropathy does not accumulate [111In]leukocytes. We found the combined bone scan and [111In] leukocyte study useful for the detection and localization of infection to soft tissue or bone in patients with neuropathic foot disease.

In vivo kinetics of canine leukocytes labeled with technetium-99m HM-PAO and indium-111 tropolonate.

Two weeks after the introduction of osteomyelitis in three dogs, autologous leukocytes were dual-labeled with both [99mTc]HM-PAO and [111In]tropolonate, and reinjected. Blood sampling and imaging were then performed. Two weeks later, the same dogs received simultaneous injections of singly-labeled [99mTc]WBC and [111In]WBC for comparison. For both studies, blood samples were drawn over 6 hr to determine the respective blood clearance half-time (TB) and % recovery (%R0) of cell-bound radioactivity. There were no significant differences in the average TB results of the 99mTc and 111In groups, either within or between the dual- and singly-labeled studies. The %R0 of singly-labeled [99mTc]WBC was about half that of the other groups (p less than 0.01); however, this difference was attributed to the dissimilar radiochemical purity of the [99mTc]HM-PAO reagents. Region of interest analysis of the 6 and 24 hr images revealed no significant differences between either cell label in the relative or absolute in vivo uptake at known sites of osteomyelitis, noninfected surgery, and normal bone marrow.