A maximum likelihood approach to continuous speech recognition.

Speech recognition is formulated as a problem of maximum likelihood decoding. This formulation requires statistical models of the speech production process. In this paper, we describe a number of statistical models for use in speech recognition. We give special attention to determining the parameters for such models from sparse data. We also describe two decoding methods, one appropriate for constrained artificial languages and one appropriate for more realistic decoding tasks. To illustrate the usefulness of the methods described, we review a number of decoding results that have been obtained with them.

Study of serum zinc in neonates and their mothers in Shimla hills (Himachal Pradesh).

Serum zinc level in cord blood of 159 neonates was estimated by atomic absorption spectrophotometer. The cases were classified according to birth weight and gestation of babies as Term appropriate for date (TAFD), Term small for date (TSFD), Term large for date (TLFD), Preterm appropriate for date (PAFD), Preterm small for date (PSFD) and Preterm large for date (PLFD). The zinc level were also estimated in mothers of these groups at the time of delivery, and compared with cord blood levels of those in non-pregnant mothers. Mean serum zinc level in infant born full term AFD, full term SFD, full term LFD, preterm AFD, preterm SFD and preterm LFD were 79.6 +/- 17.8 micrograms/dl, 58.2 +/- 13.4 micrograms/dl, 84.1 +/- 21.1 micrograms/dl, 81 +/- 25.2 micrograms/dl, 51.2 +/- 51.7 micrograms/dl and 76 +/- 14.7 micrograms/dl respectively. The maternal zinc levels in respective groups were 67 +/- 9.6 micrograms/dl, 56.5 +/- 7.5 micrograms/dl, 63.6 +/- 14.4 micrograms/dl, 62.7 +/- 21.1 micrograms/dl, 54.5 +/- 5.4 micrograms/dl, and 58.2 +/- 2.7 micrograms/dl. The mean serum zinc values in mothers and babies in birth weight group ranging from 1500-2000 gm were 55.3 +/- 4.3 micrograms/dl and 60 +/- 23.1 micrograms/dl, 2001-2500 gm were 59.5 +/- 11.3 and 65.8 +/- 17 micrograms/dl, 2501-3000 gm were 69.2 +/- 9.5 and 84.7 +/- 14 micrograms/dl, 3001-3500 gm were 65.8 +/- 12.7 micrograms/dl, 82.2 +/- 20.8 micrograms/dl and 3501 and above were 70.5 +/- 8.2 micrograms/dl and 85 +/- 14.3 micrograms/dl respectively. Statistically significant low zinc levels were observed in SFD babies and their mothers.(ABSTRACT TRUNCATED AT 250 WORDS)

Childhood trichobezoar.

Two cases of trichobezoar with unusual presentation in rural female children are described. The first one in a healthy asymptomatic child with no abnormal psychological behaviour and other one in an emotionally disturbed child with history of trichotillomania. Both were treated surgically with no recurrence. A physical sign of indentibility is discussed and literature is also reviewed.

Poisoning in children: Indian scenario.

The retrospective data on childhood poisoning from eight regional hospitals in India has been reviewed. The demographic features and types of poisonings encountered have been compared. The analysis of the data indicated that pediatric poisonings constituted 0.23-3.3% of the total poisoning. The mortality ranged from 0.64-11.6% with highest being from Shimla. Accidental poisoning was common involving 50-90% of children below 5 years of age and males outnumbered the females. Suicidal poisoning was seen after 13 years of age and was due to drugs and household chemicals. One of the hospitals in Delhi recorded a very high incidence (66.6%) of drug poisoning in children. The drugs consumed belonged to phenothiazines, antiepileptics and antipyretics. Iron poisoning was seen in younger children. Kerosene was one of the causes of accidental poisoning at all hospitals except Shimla and rural Maharashtra were probably wood charcoal is widely used. Pesticide poisoning was more prevalent in Punjab and West Bengal whereas plant poisoning was very common in Shimla. Significant number of snake envenomation has been recorded from rural Maharashtra. Other less common accidental poisonings in children included alcohol, corrosives, heavy metals, rodenticides, detergents and disinfectants. Thus various regions in the country showed some variation in types and frequency of childhood poisoning which could be attributed to different geographical and socio-economic background.

Etiology of neonatal jaundice at Shimla.

PIP: This article reports the findings of a study undertaken to evaluate the etiological factors responsible for neonatal jaundice in a high-altitude population (2000 m above sea level) in India. 164 neonates were studied, of which 105 (64%) developed clinical jaundice. 68 (64.5%) were boys and 37 (35.2%) were girls. Serum hyperbilirubinemia was found in 38 cases (23.1%). In 12 cases (11.4%), no cause could be determined for the jaundice. Physiological jaundice was determined in 67 cases (63.7%). Sepsis associated with jaundice was found in 11 cases (10.5%). This study could not establish a relationship between high altitude and the incidence of neonatal jaundice. Further studies are needed before such a relationship can be established.^ieng

Experience with diarrhea training and treatment unit in Shimla.

PIP: India's National Program for Control of Diarrheal Diseases has established Diarrhea Training and Treatment Units (DTUs) to reduce diarrhea-related morbidity and mortality. The cost-effectiveness and impact of oral rehydration therapy (ORT) were evaluated prospectively in all children presenting to the DTU of I.G. Medical College (Shimla, India) in 1993-94. During the 2-year study period, a total of 1240 children 1 month to 12 years of age attended this DTU for treatment. 47.6% were infants and 58% came from rural areas. 87% of patients had acute watery diarrhea, 10.4% had dysentery, and 2.6% had persistent diarrhea. Diarrhea was most prevalent (74.9%) from April to September. 41.9% of presenting children had received ORT before coming to the DTU; as a result, severe dehydration was seen in only 10.7% of cases. Comparison of data on 166 children admitted to the study hospital in 1986-87 (before ORT was introduced) and the 1993-94 cases revealed significant declines in the admission rate due to dehydration and/or associated illnesses (100% vs. 26.8%), use of antimicrobials (66.2% vs. 15.3%), administration of intravenous fluids (71.1% vs. 15.45%), and mortality (9.03% vs. 0.6%). 84.6% of children in 1993-94 were treated with ORT alone. The average cost of ORT per child was Rs. 4.49 compared with Rs. 40.29 for intravenous fluids. Although these findings indicate DTUs are having a favorable impact on the rational management of diarrhea, health education campaigns aimed at promoting even more widespread use of ORT by mothers and health workers are needed.^ieng

Childhood morbidity in mobile hospital camps in Himachal Pradesh.

PIP: A systematic investigation of morbidity patterns was conducted in 1977- 80 among 2580 children under 12 years of age attending mobile hospital camps in 4 districts on India's Hamachal Pradesh. The children came from remote villages where socioeconomic and educational levels were low and environmental sanitation was rudimentary. There were 1301 cases of protein energy malnutrition in this group, 124 involving children 0-1 year of age, 514 in the 1-5-year age group, and 663 (51%) in the 5-12- year age group. At the time of examination, 287 of the children were infested with worms and 125 had diarrhea. These 3 conditions-- malnutrition, worm infestation, and diarrhea--were present in 32% of the village children surveyed. The most common form of morbidity was nutritional disorders (malnutrition, anemia, and vitamin deficiencies), affecting 70% of the children. The next most common condition was respiratory infection, affecting 35%. Other disorders affecting significant numbers of children were scabies, pyoderma, convulsions, mental retardation, rheumatic fever and congenital heart diseases, and renal diseases. Morbidity from conditions such as gastroenteritis, measles, and pneumonia was often accompanied by malnutrition. Thus, there is a need in this area for child health programs aimed at providing nutrition education as well as improving immunization coverage.^ieng

Congenital cytomegalovirus infection in Shimla Hills, Himachal Pradesh, India.

We came across some cases clinically suggestive of intrauterine infection which were confirmed to be congenital cytomegalovirus infection. A clinical profile of these patients is presented. Intracranial calcification was not seen in any of these patients. Mental retardation was profound in all the patients and all had hepatomegaly. Uncommon findings encountered included hydrocephalus, patent ductus arteriosus and corneal opacities. Other clinical findings and investigation are also discussed.