George W. Dombi

Prediction of Rib Fracture Injury Outcome by an Artificial Neural Network

Outcome-based therapy is becoming the standard for assessing patient care efficacy. This study examines the ability of an artificial neural network to predict rib fracture injury outcome based on 20 intake variables determined within 1 hour of admission. The data base contained 580 patient records with four outcome variables: Length of hospital stay (LOS), ICU days, Lived, and Died. A 522-patient training set and a 58-patient test set were randomly selected. Nine networks were set up in a feed-forward, back-propagating design with each trained under different initial conditions. These networks predicted the test set outcome variables with an accuracy as high as 98% at the 80% testing level. Internal weight matrix examination indicated that age, ventilatory support, and high trauma scores were strongly associated with both ICU days and mortality. Being female, injury severity, and injury type were associated with increased LOS. Smoking and rib fracture number were low-level predictors of the four outcome variables.

Significance of micrometastases on the survival of women with T1 breast cancer

The most important factor in predicting survival among women with newly diagnosed breast cancer is the status of the axillary lymph nodes. Although straightforward to define, the impact of micrometastases on survival remains to be completely determined.

Plasma supplementation is beneficial for coagulation during severe hemorrhagic shock

BACKGROUND Hemorrhagic shock (HS) often causes coagulopathy due, in part, to decreased coagulation proteins. This study assessed the efficacy of fresh frozen plasma (FFP) in preventing this coagulopathy following a canine model of HS designed to mimic bleeding with shock as seen in the emergency department followed by bleeding without shock as seen during operation for control of bleeding. METHODS Twenty-two dogs had acute HS for 2 hours followed by resuscitation with red blood cells (RBC) plus lactated ringers (LR) or RBC and LR with FFP. After resuscitation, bleeding was continued for 1 hour while intravenous replacement of RBC and LR with or without FFP was provided. Baseline, postshock, postresuscitation, post-1 hour exchange, postoperative day one and day two measurements included coagulation Factors I, II, V, VII, VIII, and X, and the prothrombin (PT), partial thromboplastin (PTT), and thrombin times (TT). RESULTS Baseline, postshock, and postresuscitation hemodynamic responses, coagulation factor levels, and coagulation times were similar for both groups. By contrast, the 1-hour postexchange factors were depleted in the LR dogs compared to the FFP dogs. This depletion correlated with prolonged PT, PTT, and TT in the LR dogs (mean 14, 35, and 8 seconds) compared to FFP dogs (9, 24, and 6 seconds). CONCLUSIONS Severe HS beyond one blood volume exceeds the interstitial stores of coagulation protein, thus necessitating FFP supplementation.

An in vitro model of fibroblast activity and adhesion formation during flexor tendon healing

We studied fibroblast activity during tendon healing with an in vitro tendon culture model. Tendons were embedded in a translucent collagen gel matrix whose porous nature permitted free nutrient diffusion, fibroblast migration out of the tendon, and microphotographic documentation of fibroblast activity. Experiments were performed using one or more tendons cultured in the same collagen gel. We identified three zones of fibroblast activity in the gel. Zone I was an area of randomly dispersed cells directly adjacent to the tendon where collagen synthesis and remodeling were probably taking place. In zone II, spindle-shaped fibroblasts were aligned pointing away from the cut tendon end forming a sunburst-like aggregate of cells. Zone II fibroblasts were responsible for formation of migration trails by exerting a mechanical force on the collagen matrix, which was evident as a local gel contraction. Zone III was the leading edge of the sunburst populated by the fastest moving fibroblasts, which responded to guidance by other cut tendon ends. We speculate that the collagen gel used in the culture system may help maintain a chemotactic concentration gradient that allows fibroblasts to locate other distal cut tendon surfaces also embedded in the collagen gel.

Neurosurgical trauma call: Use of a mathematical simulation program to define manpower needs

Resource criteria for trauma centers (TC) mandate a first plus backup neurosurgeon (NS) coverage, an unnecessary expense for TC treating few neurosurgery patients. This report uses a mathematical modeling system to define optimal NS trauma coverage. Random data from 749 patients treated with emergency neurosurgery operations (OR) within 24 hours of admission at 97 TC were used to create a 1-year profile of admission by month, day, and hour, operation times, and operation duration. These data were entered into a simulation program to define the frequency that a patient needing a NS consult would wait beyond 30 minutes because the NS was in the operating room at a trauma center with one, two, or three neurosurgeons on-call. One thousand iterations were done for each sample size of 25 to 300 patients in 25-patient increments. The probability that a patient could not be seen promptly by one NS in a trauma center operating on 25, 50, 75, or 100 patients per year is 0.23, 0.9, 1.6, and 3.66 patients per year. Fewer than one patient (0.75) per year will wait more than 30 min in a trauma center doing 225 emergency ORs when two neurosurgeons are on-call. One patient in 10 years would wait more than 30 min in a trauma center doing 300 ORs with a third NS on-call. Mathematical modeling of patient data helps define optimal hospital resources. Mandatory NS backup for TC performing fewer than 25 neurosurgery procedures is unneeded.

Adjuvant radiation therapy increases overall survival in node-positive gastric cancer patients with aggressive surgical resection and lymph node dissection: a SEER database analysis.

ObjectivesTo determine the outcomes of postoperative radiation therapy on survival in gastric cancer. MethodsAn analysis of patients with surgically resected and nonmetastatic gastric cancer from the Surveillance, Epidemiology, and End Results database from 1990 to 2003 was carried out. Survival curves were calculated according to the Kaplan-Meier method. Multivariate analysis was carried out by the Cox proportional hazard model. ResultsWe identified 11,630 patients who met inclusion criteria. Radiation therapy was associated with increased survival in patients with American Joint Committee on Cancer stage IB to IV. The median survival for stage IB and II patients treated with radiation was 96 months and 37 months, respectively, versus 56 months and 23 months for patients who did not receive adjuvant radiation (P=0.0281 for stage IB and <0.0001 for stage II). The 5-year overall survival for node-positive patients treated with radiation was 30.4% versus 21.4% for patients who did not receive adjuvant radiation (P<0.0001). The survival benefit of radiation therapy was maintained even if ≥15 lymph nodes were removed for N1 and N2 disease and if ≥30 lymph nodes were removed for N3 disease. For node-positive patients with ≥15 lymph nodes removed, adjuvant radiation was linked to increase survival in patients who underwent partial gastrectomy, total gastrectomy, and en bloc gastrectomy with other organs removed. Radiation was a strong independent factor for survival on multivariate analysis. ConclusionsThere is a correlation between survival and radiation therapy in node-positive gastric cancer patients and is independent of the extent of surgical resection and lymph node dissection.

Improved Survival in Patients With Lymph Node-Positive Gastric Cancer Who Received Preoperative Radiation An Analysis of the Surveillance, Epidemiology, and End Results Database

Several trials have been conducted to determine the feasibility of preoperative radiotherapy (RT) for gastric cancer. However, the absolute benefit from radiotherapy remains to be defined. In this study, the authors examined the use of preoperative RT (Pre‐RT) and postoperative RT (PORT) in patients with gastric cancer from the Surveillance, Epidemiology, and End Results (SEER) database.

Neural network analysis of employment history as a risk factor for prostate cancer

BACKGROUND Prostate cancer is the most common non-cutaneous malignancy in men. Its etiology likely involves environmental exposures and demographic factors. OBJECTIVE Investigate the potential relationship between occupation history and prostate cancer risk in a population-based, case-control study (n=1365). METHODS The variables: race, age group, smoking status, income, marital status, education and the first 15 years of employment history were examined by sequential odds ratio analysis then compared to a neural network consensus model. RESULTS Both the sequential odds ratio method and the neural network consensus model identified a similar hypothetical case of greatest risk: a Black, married man, older than 60 years, with at best a high school diploma who made between

Inhibition of alpha-smooth muscle actin expression in an in vitro wound healing model by certain antibiotics

25,000-

Hemodynamic, Plasma Volume, and Prenodal Skin Lymph Responses to Varied Resuscitation Regimens

65,000. The work history determined by odds ratio analysis consisted of 10 years in the chemical industry with 3 yrs in the processing plant. Neural network analysis showed a similar work history with 8 years in the chemical industry and 2 years in the processing plant. DISCUSSION Neural network outcomes are similar to sequential odds ratio calculations. This work supported previous studies by finding well known demographic risk factors for prostate cancer including certain processing jobs and chemical related jobs.