A novel objective function for improved phoneme recognition using time-delay neural networks.

Single-speaker and multispeaker recognition results are presented for the voice-stop consonants /b,d,g/ using time-delay neural networks (TDNNs) with a number of enhancements, including a new objective function for training these networks. The new objective function, called the classification figure of merit (CFM), differs markedly from the traditional mean-squared-error (MSE) objective function and the related cross entropy (CE) objective function. Where the MSE and CE objective functions seek to minimize the difference between each output node and its ideal activation, the CFM function seeks to maximize the difference between the output activation of the node representing incorrect classifications. A simple arbitration mechanism is used with all three objective functions to achieve a median 30% reduction in the number of misclassifications when compared to TDNNs trained with the traditional MSE back-propagation objective function alone.

Growth hormone, insulin, glucose, cortisol, luteinizing hormone, and diabetes in beagle bitches treated with medroxyprogesterone acetate.

Adult beagle bitches (20 to 101 months old) received medroxyprogesterone acetate (MPA; 75 mg/kg, im) or control vehicle at 3 month intervals. Changes in serum concentrations of GH, insulin and glucose were determined in 18 MPA-treated and 6 of 12 control bitches at 0, 2, 4, 8, 16 and 17-24 months of treatment (Exp. I). GH, LH and cortisol responsiveness to combined im injection of TRH (10 micrograms/kg), GnRH (10 micrograms/kg), and ACTH (5 micrograms/kg) was determined in 9 MPA-treated and 9 control bitches at 17 months of treatment (Exp. II). In Exp. I, serum concentrations of GH at month 2 (2.6 +/- 0.3 micrograms/l), 4 (3.0 +/- 0.3 micrograms/l), 8 (4.0 +/- 1.2 micrograms/l), 16 (8.5 +/- 1.7 micrograms/l), and 17-24 (21.2 +/- 4.1 micrograms/l) of treatment were greater (P less than 0.05) than pretreatment (1.4 +/- 0.07 micrograms/l) and control (1.5 +/- 0.1 microgram/l) levels. The increase in GH at 2 months preceded (N = 4) or coincided (N = 2) with the development of hyperinsulinaemia and insulin resistance in 6 of the 18 treated bitches, two of which became diabetic by 17 months of MPA treatment. GH (24.6 +/- 5.0 vs 11.4 +/- 2.1 micrograms/l) and insulin (308 +/- 77 vs 119 +/- 9 pmol/l) concentrations were greater (P less than 0.05) in older (49 +/- 4 months; N = 12) than in the younger (26 +/- 2 months; N = 6) treated bitches at 17-24 months of MPA treatment. In Exp. II, pretreatment concentrations of GH were increased (9.8 +/- 3.0 vs 1.4 +/- 0.1 micrograms/l, P less than 0.01).(ABSTRACT TRUNCATED AT 250 WORDS)

Intranasal instillation of oxytocin increases insulin and glucagon secretion.

Intravenously administered oxytocin was found to increase plasma insulin and glucagon levels. To explore if the same effects could be obtained by nonparenteral routes of administration, oxytocin was given by nasal instillation in normal conscious dogs. Plasma glucose, insulin, and glucagon levels all increased to levels which previously were shown to cause increased glucose production and utilization. Vasopressin infusion had no effect on these measurements. This is the first report of the effectiveness of oxytocin to evoke insulin and glucagon secretion by the nasal route of administration.

Clonidine or xylazine as provocative tests for growth hormone secretion in the dog.

Base-line values of plasma growth hormone (GH) are low in most species, requiring provocative tests to assess GH deficiency. Clonidine, an antihypertensive drug, and its analogue, xylazine, a sedative hypnotic, were found to stimulate GH secretion. Administration (IV) of clonidine to conscious healthy, dogs at doses of 30, 16.5, and 3 microgram/kg produced significant increases in plasma GH by 15 minutes and the effects subsided by 120 minutes. Plasma glucose concentration increased slowly with all doses, but less so and for shorter duration in dogs given the 3 microgram/kg dose. Xylazine increased plasma GH when injected at doses of 300 and 100 microgram/kg, but not at 30 microgram/kg. Plasma glucose increased only with the 300 microgram/kg dose. The alpha-adrenergic blocker, phentolamine, markedly attenuated these responses. Thus, both clonidine and xylazine, when used at appropriate doses, can stimulate GH secretion, with minimal effect on plasma glucose and without causing significant sedation.

Oxytocin infusion increases plasma insulin and glucagon levels and glucose production and uptake in the normal dog.

Infusion of oxytocin (50--500 microU/kg/min) into normal conscious dogs produces a rise in plasma glucose, insulin, and glucagon levels. These changes are accompanied by a prompt increase in glucose production followed by an increase in overall glucose uptake, as determined using 6-3H-glucose infusion.

Acromegaly associated with transient overproduction of growth hormone in a dog.

A 6-year-old female crossbred Belgian Shepherd Dog with features of acromegaly was monitored for almost 4 years. The history of frequent and excessive administration of a progestational agent suggested that the progestational drug induced the acromegaly. During the monitoring period. soft tissue changes diminished and there was normalization of several factors: plasma growth hormone concentration, response of plasma insulin and glucose to an oral glucose load, and response of plasma glucose hormone to the injection of insulin.

Thymic abnormalities and growth hormone deficiency in dogs.

A high frequency of occurrence of a wasting disease, unthriftiness, and retarded growth was observed in a group of inbred Weimaraner dogs. Affected pups had a small thymus gland, with a marked absence of thymic cortex. A litter of eight pups from a sire and dam that were known to have produced affected offspring was chosen for further study. The pups had normal concentrations of WBC and gamma-globulins and were able to produce antibody in response to Brucella abortus. Two pups in the litter developed a wasting syndrome and responded well to therapy with thymosin fraction 5. One pup that survived the wasting syndrome had a significant (P < 0.05) depression of its lymphocyte blastogenic response to phytohemagglutinin compared with its surviving littermates. Pups from this litter also lacked a normal increase in plasma growth hormone concentration after the injection of clonidine HCl. These pups had concurrent abnormalities of the thymus-dependent immune function and in growth hormone metabolism. The syndrome in these pups has some features in common with the syndrome in the Ames or Snell-Bagg strains of immunodeficient dwarf mice.

Grwoth hormone, prolactin, and cortisol in dogs developing mammary nodules and an acromegaly-like appearance during treatment with medroxyprogesterone acetate.

PIP: Concentrations of (GH) growth hormone, (PRI) prolactin, cortisol, progesterone, and (MPA) medroxyprogesterone acetate were determined by RIA in blood sera collected from beagle bitches 17 months after initiating treatment with MPA (75 mg/kg.3 months; n = 12) MPA vehicle (controls; n = 12), or progesterone implants which produced physiological levels of progesterone (13.8 + or - 2.1 ng/ml; n = 12). In the MPA-treated bitches, mean MPA levels were 104 + or - 6 ng/ml, mean GH levels were elevated (9.5 + or - 3.0 vs. 0.4 + or - 0.1 ng/ml; P 0.01); mean PRL levels were unchanged (13.7 + or - 2.8 vs. 12.6 + or - 1.2 ng/ml); and mean cortisol levels were suppressed (1.7 + or - 0.2 vs. 13.7 + or - 1.4 ng/ml; P 0.01) in comparison to those in control animals. None of these parameters was significantly affected by progesterone treatment. External signs of an acromegaly-like condition and large mammary gland nodules (diameters, 5 mm) were noted in, and limited to, 9 bitches with elevated ( 2.5 ng/ml) GH levels (12.8 + or - 3.0 ng/ml). These were 8 MPA-treated bitches which developed the acromegal-like condition during treatment and 1 progesterone-treated bitch which appeared acromegalic before treatment and in which the condition was considered to have developed spontaneously. The data suggest that the acromegaly-like changes and large mammary nodules in dogs administered the contraceptive progestin MPA occurred as a result of MPA-induced elevations in GH. The results do not preclude the possibility that the MPA-induced suppression of cortisol and/or the direct action of MPA on the mammary glands also contributed to mammary nodule formation. MPA-treated dogs may also provide a unique experimental model for studying chronic elevations in endogenous GH levels and for testing compounds for their ability to suppress GH levels.^ieng

On the mechanism of hyperglycemia and stimulation of growth hormone secretion by L-dopa.

The mechanism of hyperglycemia produced by L-dopa was studied in normal trained dogs with 3-3H-glucose infusion to measure rates of hepatic glucose output (production) and overall glucose uptake (utilization). Infusion of L-dopa (20 mg/kg/h) increased glucose production causing hyperglycemia. Despite the hyperglycemia plasma insulin did not increase nor did glucose uptake, indicating a relative inhibition of glucose utilization. These effects resemble those produced by epinephrine infusion. Pretreatment with a decarboxylase inhibitor, carbidopa, prevented the L-dopa effect to increase glucose production and no hyperglycemia occurred. Hyperglycemia was not prevented by pimozide, a dopamine receptor blocker, nor by propranolol but was prevented by phentolamine. L-Dopa also increased plasma growth hormone levels without affecting plasma cortisol. The effect on growth hormone was prevented by carbidopa and by phentolamine but not by pimozide; propranolol potentiated the rise in growth hormone. The data suggest that the L-dopa-induced hyperglycemia is due to a peripheral action, whereas stimulation of growth hormone secretion may be due to a central action of a L-dopa metabolite.

On the mechanism of diazoxide-induced hyperglycemia.

Infusion of diazoxide (16.5 mg./kg. in 10 minutes) into normal unanesthetized dogs resulted in a prompt hyperglycemia due to increased hepatic glucose production as measured with a 3-3H-glucose primer-infusion technique. Plasma insulin and glucagon were decreased. Glucose uptake failed to increase. Diazoxide administration during period of alpha adrenergic receptor blockade with phentolamine still caused hyperglycemia and increased glucose production. Glucose uptake was inhibited despite adequate plasma insulin. Infusion of somatostatin along with insulin prevented the effects of diazoxide on plasma glucose and glucose production. It is concluded that diazoxide hyperglycemia is not due solely to decreased insulin secretion or increased epinephrine secretion and that glucagon is not a contributory factor. Diazoxide may act directly to increase glucose production and inhibit glucose uptake. Somatostatin appears capable of blocking the effect of diazoxide on glucose production by an unknown mechanism.

Interaction of somatostatin, glucagon, and insulin on hepatic glucose output in the normal dog.

The interaction of insulin and glucagon during infusion of somatostatin (SRIF), which suppresses secretion of these hormones, was investigated in normal, postabsorptive, concious dogs. Hepatic glucose output (production) and over-all glucose uptake by the tissues was measured with 3-3H-glucose, administered by a priming injection along with a constant infusion. Infusion of SRIF (1.5-5.0 mug/min) for 90 minutes resulted in a moderate hypoglycemia associated with a decrease in glucose production. In some animals glucose production and plasma glucose levels returned to normal before the end of SRIF infusion. Glucose uptake tended to follow plasma glucose levels. Upon termination of SRIF infusion, glucose production and uptake and plasma glucose increased sharply.

Radioimmunoassay of canine growth hormone: enzymatic radioiodination.

A sensitive radioimmunoassay is described for the measurement of plasma concentrations of canine growth hormone (cGH) as low as 0.25 ng/ml. The assay utilizes enzymatically iodinated cGH and the double antibody technique. The mean plasma concentration of growth hormone in the normal dog after overnight fast is 1.75 plus or minus .17 ng/ml. Exogenous cGH was cleared from the plasma of both the normal and hypophysectomized dog with a mean half-life of 25.6 plus or minus 1.0 min and was distributed in a volume equal to 8.9% of the body weight. Insulin hypoglycemia produced a 3- to 5-fold increase in plasma GH in 4 of 6 dogs and arginine infusion failed to produce a statistically significant rise.