ABSTRACT
Objective:To investigate the prenatal ultrasonographic features and diagnosis of 16p12.2 copy number variation (CNV).Methods:This retrospective study recruited seven fetuses with 16p12.2 microdeletion/microduplication in the First Affiliated Hospital of Fujian Medical University from January 2017 to December 2021. Data, including the prenatal diagnostic indications, ultrasound findings, karyotypes, genetic testing and mutation tracing results, pregnancy outcomes, and postnatal follow-up data, were summarized with descriptive statistical analysis.Results:Prenatal ultrasound indicated three fetuses with structural abnormalities, including one case each of multiple malformations, interventricular septal defect, and cleft lip and palate. The other four cases were positive for ultrasonic soft markers involving the heart and kidney. The chromosome karyotypes of the seven fetuses were normal. Single nucleotide polymorphism array (SNP array) results showed that four cases had a 381.7-542.4 kb microdeletion containing three genes ( OTOA, METTL9, and IGSF6) in Online Mendelian Inheritance in Man (OMIM) at 16p12.2 (distal region) and three cases had a 484.0-701.7 kb microdeletion/microduplication containing four OMIM genes ( UQCRC2, CDR2, EEF2K, and POLR3E) at 16p12.2 (proximal region). Five (cases 1, 2, 4, 5, and 6) out of the seven fetuses inherited the variants from their phenotypically normal mother/father, and among them, three (cases 2, 4, and 5) were delivered at term and healthy. Two cases (cases 3 and 7) refused to undergo pedigree verification. Case 3, a full-term infant, underwent ventricular septal defect repair three months after birth, and no abnormality was found at 18 months of age. Conclusions:No specific phenotype presents in fetuses with 16p12.2 microdeletion/microduplication in prenatal diagnosis. OTOA gene is the key gene associated with abnormality in the distal region of 16p12.2. Pedigree analysis is conducive to preventing unnecessary termination of pregnancy.
ABSTRACT
In this paper, we present an effective deep prediction framework based on robust recurrent neural networks (RNNs) to predict the likely therapeutic classes of medications a patient is taking, given a sequence of diagnostic billing codes in their record. Accurately capturing the list of medications currently taken by a given patient is extremely challenging due to undefined errors and omissions. We present a general robust framework that explicitly models the possible contamination through overtime decay mechanism on the input billing codes and noise injection into the recurrent hidden states, respectively. By doing this, billing codes are reformulated into its temporal patterns with decay rates on each medical variable, and the hidden states of RNNs are regularized by random noises which serve as dropout to improved RNNs robustness towards data variability in terms of missing values and multiple errors. The proposed method is extensively evaluated on real health care data to demonstrate its effectiveness in suggesting medication orders from contaminated values.
