RESUMO
Health and disease are fundamentally influenced by microbial communities and their genes (the microbiome). An in-depth analysis of microbiome structure that enables the classification of individuals based on their health can be crucial in enhancing diagnostics and treatment strategies to improve the overall well-being of an individual. In this paper, we present a novel semi-supervised methodology known as Randomized Feature Selection based Latent Dirichlet Allocation (RFSLDA) to study the impact of the gut microbiome on a subject's health status. Since the data in our study consists of fuzzy health labels, which are self-reported, traditional supervised learning approaches may not be suitable. As a first step, based on the similarity between documents in text analysis and gut-microbiome data, we employ Latent Dirichlet Allocation (LDA), a topic modeling approach which uses microbiome counts as features to group subjects into relatively homogeneous clusters, without invoking any knowledge of observed health status (labels) of subjects. We then leverage information from the observed health status of subjects to associate these clusters with the most similar health status making it a semi-supervised approach. Finally, a feature selection technique is incorporated into the model to improve the overall classification performance. The proposed method provides a semi-supervised topic modelling approach that can help handle the high dimensionality of the microbiome data in association studies. Our experiments reveal that our semi-supervised classification algorithm is effective and efficient in terms of high classification accuracy compared to popular supervised learning approaches like SVM and multinomial logistic model. The RFSLDA framework is attractive because it (i) enhances clustering accuracy by identifying key bacteria types as indicators of health status, (ii) identifies key bacteria types within each group based on estimates of the proportion of bacteria types within the groups, and (iii) computes a measure of within-group similarity to identify highly similar subjects in terms of their health status.
Assuntos
Microbioma Gastrointestinal , Microbiota , Humanos , AlgoritmosRESUMO
Among the most prevalent neurodevelopmental disorders, Autism Spectrum Disorder (ASD) is highly diverse showing a broad phenotypic spectrum. ASD also couples with a broad range of mutations, both de novo and inherited. In this study, we used a proprietary SNP genotyping chip to analyze the genomic DNA of 250 Vietnamese children diagnosed with ASD. Our Single Nucleotide Polymorphism (SNP) genotyping chip directly targets more than 800 thousand SNPs in the genome. Our primary focus was to identify pathogenic/likely pathogenic mutations that are potentially linked to more severe symptoms of autism. We identified and validated 23 pathogenic/likely pathogenic mutations in this initial study. The data shows that these mutations were detected in several cases spanning multiple biological pathways. Among the confirmed SNPs, mutations were identified in genes previously known to be strongly associated with ASD such as SLCO1B1, ACADSB, TCF4, HCP5, MOCOS, SRD5A2, MCCC2, DCC, and PRKN while several other mutations are known to associate with autistic traits or other neurodevelopmental disorders. Some mutations were found in multiple patients and some patients carried multiple pathogenic/likely pathogenic mutations. These findings contribute to the identification of potential targets for therapeutic solutions in what is considered a genetically heterogeneous neurodevelopmental disorder.
Assuntos
Transtorno do Espectro Autista , Criança , Humanos , Transtorno do Espectro Autista/genética , Polimorfismo de Nucleotídeo Único , Genótipo , Vietnã , Predisposição Genética para Doença , Mutação , Transportador 1 de Ânion Orgânico Específico do Fígado/genética , Sulfurtransferases/genética , Proteínas de Membrana/genética , 3-Oxo-5-alfa-Esteroide 4-Desidrogenase/genéticaRESUMO
ABSTRACT Objective: To compare radiographic and surgical outcomes of Lenke 1B and 1C patterns. Methods: One hundred twenty patients with Lenke 1B and 1C scoliosis were grouped according to implant density as follows: low density (LD) of ≤1.4 and high density (HD) of >1.4. Matched subgroups (30 patients each) based on age, curve magnitude, and body mass index (BMI) were analyzed. Radiographic parameters were evaluated before operation, immediately after operation (ipo), and at 2 years' follow-up. SRS-30 was administered before operation and at 2 years' follow-up. Results: The major curves of the LD (n = 82) and HD groups (n=38) were respectively 59.1° and 65.6° before operation (p <.001), 26.3° and 22.9° ipo (p =.05), and 29.9° and 19.8° at 2 years' follow-up (p <.001). No significant differences in postoperative trunk shift and coronal balance were found (p =.69 and p =.74, respectively). The HD group had higher blood loss (p =.02), number of implants (p <.001), levels fused (p =.002), and surgical time (p <.001). The HD group had a higher prevalence of hypokyphosis from before operation to follow-up (p <.001). No significant differences were observed in the SRS-30 scores before operation and at 2 years' follow-up. The matched groups had similar preoperative major curves (p =.56), ages (p =.75), and BMIs (p =.61). Significantly longer surgical time (p =.009), higher density (p <.001), and better correction (p =.0001) were found in the HD group at 2 years' follow-up. No significant differences were found in the SRS-30 scores before operation and at 2 years' follow-up. Conclusion: LD constructs included fewer segments fused, lower intraoperative estimated surgical blood loss, and shorter operation time, and potentially decreasing complication risks due to fewer implants. Level of evidence III, Retrospective Cohort Study.
RESUMO Objetivo: Comparar os desfechos radiográficos e cirúrgicos da escoliose Lenke 1B e 1C. Métodos: Cento e vinte pacientes com escoliose Lenke 1B e 1C foram agrupados de acordo com a densidade do implante, como segue: baixa densidade (BD) de ≤ 1,4 e alta densidade (AD) de > 1,4. Foram analisados os grupos pareados (30 pacientes cada) com base na idade, magnitude da curva e índice de massa corporal (IMC). Os parâmetros radiográficos foram avaliados antes da cirurgia, no pós-operatório imediato (POI) e no acompanhamento de dois anos. O questionário SRS-30 foi administrado antes da cirurgia e no acompanhamento de dois anos. Resultados: As principais curvas dos grupos BD (n = 82) e AD (n = 38) foram respectivamente 59,1° e 65,6° antes da operação (p < 0,001), 26,3° e 22,9° no POI (p = 0,05) e 29,9° e 19,8° aos 2 anos de acompanhamento (p < 0,001). Não foram encontradas diferenças significantes no desvio do tronco e no balanço coronal no pós-operatório (p = 0,69 e p = 0,74, respectivamente). O grupo AD teve mais perda sanguínea (p = 0,02), número de implantes (p < 0,001), níveis de fusão (p = 0,002) e tempo de cirurgia (p < 0,001). O grupo AD teve maior prevalência de hipocifose do período anterior à cirurgia até o acompanhamento (p < 0,001). Não houve diferenças significantes nas pontuações do SRS-30 antes da operação e aos 2 anos de acompanhamento. No pré-operatório, os grupos pareados tinham curvas principais (p = 0,56), idade (p = 0,75) e IMC (p = 0,61) semelhantes. Constatou-se tempo cirúrgico expressivamente maior (p = 0,009), maior densidade (p < 0,001) e melhor correção (p = 0,0001) no grupo AD aos 2 anos de acompanhamento. Não foram encontradas diferenças significantes nas pontuações do SRS-30 antes da cirurgia e no acompanhamento de 2 anos. Conclusão: As estruturas de BD incluíram menos segmentos fundidos, menor perda de sangue intraoperatória estimada, menor tempo de cirurgia e menos risco de complicações, com possibilidade de redução, por causa do menor número de implantes. Nível de evidência III, Estudo retrospectivo de coorte.