Clarifying Ambiguous Keywords with Personal Word Embeddings for Personalized Search

Personalized search tailors document ranking lists for each individual user based on her interests and query intent to better satisfy the user’s information need. Many personalized search models have been proposed. They first build a user interest profile from the user’s search history, and then re-rank the documents based on the personalized matching scores between the created profile and candidate documents. In this article, we attempt to solve the personalized search problem from an alternative perspective of clarifying the user’s intention of the current query. We know that there are many ambiguous words in natural language such as “Apple.” People with different knowledge backgrounds and interests have personalized understandings of these words. Therefore, we propose a personalized search model with personal word embeddings for each individual user that mainly contain the word meanings that the user already knows and can reflect the user interests. To learn great personal word embeddings, we design a pre-training model that captures both the textual information of the query log and the information about user interests contained in the click-through data represented as a graph structure. With personal word embeddings, we obtain the personalized word and context-aware representations of the query and documents. Furthermore, we also employ the current session as the short-term search context to dynamically disambiguate the current query. Finally, we use a matching model to calculate the matching score between the personalized query and document representations for ranking. Experimental results on two large-scale query logs show that our designed model significantly outperforms state-of-the-art personalization models.

First Results From a Propensity Matching Trial of Mycophenolate Mofetil vs. Azathioprine in Treatment-Naive AIH Patients

Background/AimsAs previous real-world studies and meta-analyses have shown that mycophenolate mofetil (MMF) might have better efficacy than azathioprine (AZA) in autoimmune hepatitis (AIH), we conducted a propensity matching study to assess the efficacy and safety of MMF vs. AZA.MethodsAll 126 consecutive treatment-naive adult AIH patients, diagnosed and followed in our department since 2016, were included. Patients received prednisolone 0.5–1 mg/kg/day plus either AZA 1–2 mg/kg/day or 1.5–2 g/day MMF. The tapering of prednisolone was identical between groups.ResultsAfter propensity matching score and adjustment for known factors affecting response to treatment and outcome, 64 patients were included in the study (MMF = 32 and AZA = 32). Rates of non-response, complete biochemical response (CBR) at 6 and 12 months, and prednisolone withdrawal (6 months, 12 months, and end of follow-up) were identical between groups. However, MMF treatment was significantly associated with CBR at the end of follow-up [odds ratio (OR) 11.259; 95% CI: 1.3–97.4, p = 0.028]. AZA patients were more prone to stop treatment due to AZA intolerance/insufficient response (p = 0.0001). At the end of follow-up, the overall efficacy of each schedule was also significantly higher in the MMF group compared to the AZA group (p = 0.0001).ConclusionWe showed for the first time in a propensity matching study that MMF can be used as first-line therapy in AIH as attested by the significantly higher CBR at end of follow-up compared to AZA. Whether this better efficacy is also associated with higher histological remission rates and sustained CBR off immunosuppression needs further evaluation.

Implementing a protocol to prevent incisional hernia in high-risk patients: a mesh is a powerful tool

Purpose The small bites (SB) technique for closure of elective midline laparotomies (EMLs) and a prophylactic mesh (PM) in high-risk patients are suggested by the guidelines to prevent incisional hernias (IHs) and fascial dehiscence (FD). Our aim was to implement a protocol combining both the techniques and to analyze its outcomes. Methods Prospective data of all EMLs were collected for 2 years. Results were analyzed at 1 month and during follow-up. The incidence of HI and FD was compared by groups (M = Mesh vs. S = suture) and by subgroups depending on using SB. Results A lower number of FD appeared in the M group (OR 0.0692; 95% CI 0.008–0.56; P = 0.01) in 197 operations. After a mean follow-up of 29.23 months (N = 163; min. 6 months), with a lower frequency of IH in M group (OR 0.769; 95% CI 0.65–0.91; P < 0.0001). (33) The observed differences persisted after a propensity matching score: FD (OR 0.355; 95% CI 0.255–0.494; P < 0.0001) and IH (OR 0.394; 95% CI 0.24–0.61; P < 0.0001). On comparing suturing techniques by subgroups, both mesh subgroups had better outcomes. PM was the main factor related to the reduction of IH (HR 11.794; 95% CI 4.29–32.39; P < 0.0001). Conclusion Following the protocol using PM and SB showed a lower rate of FD and HI. A PM is safe and effective for the prevention of both HI and FD after MLE, regardless of the closure technique used.

Therapeutic Actionability of Circulating Cell-Free DNA Alterations in Carcinoma of Unknown Primary

PURPOSE Cancer of unknown primary (CUP) is a metastatic disease with unidentifiable primary tumor. Somatic alterations can be assessed noninvasively via liquid biopsies interrogating cell-free DNA (cfDNA). METHODS We evaluated 1,931 patients with CUP with a cfDNA next-generation sequencing panel (73-74 genes). RESULTS Overall, 1,739 patients (90%) had ≥ 1 cfDNA alteration. We then explored alteration actionability (per the levels of evidence from the OncoKB database); 825 patients (47.4% of 1,739) had level 1, level 2, or resistance/R1 alterations. Among 40 clinically annotated patients with CUP who had cfDNA evaluated, higher degrees of matching treatment to alterations (Matching Score > 50% v ≤ 50%) was the only variable predicting improved outcome: longer median progression-free survival (10.4 v 2.5 months; P = .002), overall survival (13.4 v 5.7 months; P = .07, trend), and higher clinical benefit rate (stable disease ≥ 6 months/partial response/complete response; 83% v 25%; P = .003). CONCLUSION In summary, cfDNA frequently reveals strong level-of-evidence actionable alterations in CUP, and high degrees of matching to therapy correlates with better outcomes.

O03 SMALL BITES AND PROPHYLACTIC MESH TO PREVENT INCISIONAL HERNIA IN HIGH-RISK PATIENTS. A MESH IS THE MOST POWERFUL TOOL

Aim Small bites (SB) technique for closure of elective midline laparotomies (EMLs) and a prophylactic mesh (PM) in high-risk patients are suggested by guidelines to prevent incisional hernias (IHs) and burst abdomen (BA). Our aim was to implement a protocol combining both and to analyze its outcomes. Material and Methods Prospective data collection of all EMLs for one year. Results were analyzed at one month and during follow-up. The incidence of IH and BA was compared by groups depending on the use of a PM (M Group) or not (S Group), and in subgroups related to the closure technique: SB (Subgroups MSB and SSB) or not (Subgroups MLB and MSB). Results A lower number of BA was diagnosed in the M group (OR 0.0692; CI95% 0.008-0.56; P = 0.01) in 197 operations. 163 patients completed a mean follow-up of 29.23 months, with a lower frequency of IH in M group patients (OR = 0.769; CI 95% 0.65 - 0.91; P &lt; 0.0001). Same differences persisted after a propensity matching score: BA (OR = 0.355; CI 95% 0.255 - 0.494; P &lt; 0.0001) and IH (OR = 0.394; CI 95% 0.24 - 0.61; P &lt; 0.0001). Comparing suturing techniques by subgroups any difference in IH and BA appeared. PM was the main factor related to reduction of IH (HR 11.794; CI 95% 4.29 - 32.39; P &lt; 0.0001). Conclusions A PM is the most powerful tool for prevention both IH and BA after EMLs, regardless of the closure technique in patients at high-risk for IHs.

Integrating Vision System to a Pick and Place Cartesian Robot

Vision aided pick and place cartesian robot is a combination of machine vision system and robotic system. They communicate with each other simultaneously to perform object sorting. In this project, machine vision algorithm for object sorting to solve the problem in failure sorting due to imperfection of images edges and different types of colours is proposed. The image is acquired by a camera and followed by image calibration. Pre-processing of image is performed through these methods, which are HSI colour space transformation, Gaussian filter for image filtering, Otsu’s method for image binarization, and Canny edge detection. LabVIEW edge-based geometric matching is selected for template matching. After the vision application analysed the image, electrical signal will send to robotic arm for object sorting if the acquired image is matched with template image. The proposed machine vision algorithm has yielded an accurate template matching score from 800 to 1000 under different disturbances and conditions. This machine vision algorithm provides more customizable parameters for each methods yet improves the accuracy of template matching.

A Multi-Marker Test for Analyzing Paired Genetic Data in Transplantation

Emerging evidence suggests that donor/recipient matching in non-HLA (human leukocyte antigen) regions of the genome may impact transplant outcomes and recognizing these matching effects may increase the power of transplant genetics studies. Most available matching scores account for either single-nucleotide polymorphism (SNP) matching only or sum these SNP matching scores across multiple gene-coding regions, which makes it challenging to interpret the association findings. We propose a multi-marker Joint Score Test (JST) to jointly test for association between recipient genotype SNP effects and a gene-based matching score with transplant outcomes. This method utilizes Eigen decomposition as a dimension reduction technique to potentially increase statistical power by decreasing the degrees of freedom for the test. In addition, JST allows for the matching effect and the recipient genotype effect to follow different biological mechanisms, which is not the case for other multi-marker methods. Extensive simulation studies show that JST is competitive when compared with existing methods, such as the sequence kernel association test (SKAT), especially under scenarios where associated SNPs are in low linkage disequilibrium with non-associated SNPs or in gene regions containing a large number of SNPs. Applying the method to paired donor/recipient genetic data from kidney transplant studies yields various gene regions that are potentially associated with incidence of acute rejection after transplant.

Molecular profiling of advanced malignancies guides first-line N-of-1 treatments in the I-PREDICT treatment-naïve study

Background Malignancies are molecularly complex and become more resistant with each line of therapy. We hypothesized that offering matched, individualized combination therapies to patients with treatment-naïve, advanced cancers would be feasible and efficacious. Patients with newly diagnosed unresectable/metastatic, poor-prognosis cancers were enrolled in a cross-institutional prospective study. Methods A total of 145 patients were included in the study. Genomic profiling (tissue and/or circulating tumor DNA) was performed in all patients, and PD-L1 immunohistochemistry, tumor mutational burden, and microsatellite status assessment were performed in a subset of patients. We evaluated safety and outcomes: disease-control rate (stable disease for ≥ 6 months or partial or complete response), progression-free survival (PFS), and overall survival (OS). Results Seventy-six of 145 patients (52%) were treated, most commonly for non-colorectal gastrointestinal cancers, carcinomas of unknown primary, and hepatobiliary malignancies (53% women; median age, 63 years). The median number of deleterious genomic alterations per patient was 5 (range, 0–15). Fifty-four treated patients (71%) received ≥ 1 molecularly matched therapy, demonstrating the feasibility of administering molecularly matched therapy. The Matching Score, which reflects the percentage of targeted alterations, correlated linearly with progression-free survival (R2 = 0.92; P = 0.01), and high (≥ 60%) Matching Score was an independent predictor of improved disease control rate [OR 3.31 (95% CI 1.01–10.83), P = 0.048], PFS [HR 0.55 (0.28–1.07), P = 0.08], and OS [HR 0.42 (0.21–0.85), P = 0.02]. Serious adverse event rates were similar in the unmatched and matched groups. Conclusions Personalized combination therapies targeting a majority of a patient’s molecular alterations have antitumor activity as first-line treatment. These findings underscore the feasibility and importance of using tailored N-of-1 combination therapies early in the course of lethal malignancies. Trial registration I-PREDICT (NCT02534675) was registered on August 25, 2015.

Enhance the Language Ability of Humanoid Robot NAO through Deep Learning to Interact with Autistic Children

Autism spectrum disorder (ASD) is a life-long neurological disability, and a cure has not yet been found. ASD begins early in childhood and lasts throughout a person’s life. Through early intervention, many actions can be taken to improve the quality of life of children. Robots are one of the best choices for accompanying children with autism. However, for most robots, the dialogue system uses traditional techniques to produce responses. Robots cannot produce meaningful answers when the conversations have not been recorded in a database. The main contribution of our work is the incorporation of a conversation model into an actual robot system for supporting children with autism. We present the use a neural network model as the generative conversational agent, which aimed at generating meaningful and coherent dialogue responses given the dialogue history. The proposed model shares an embedding layer between the encoding and decoding processes through adoption. The model is different from the canonical Seq2Seq model in which the encoder output is used only to set-up the initial state of the decoder to avoid favoring short and unconditional responses with high prior probability. In order to improve the sensitivity to context, we changed the input method of the model to better adapt to the utterances of children with autism. We adopted transfer learning to make the proposed model learn the characteristics of dialogue with autistic children and to solve the problem of the insufficient corpus of dialogue. Experiments showed that the proposed method was superior to the canonical Seq2sSeq model and the GAN-based dialogue model in both automatic evaluation indicators and human evaluation, including pushing the BLEU precision to 0.23, the greedy matching score to 0.69, the embedding average score to 0.82, the vector extrema score to 0.55, the skip-thought score to 0.65, the KL divergence score to 5.73, and the EMD score to 12.21.

A Study on the Diagnostic Abilities of Ultrasound Scans in Assessing the Uterine Fibroids Against Magnetic Resonance Imaging Findings in the Same Subject Population

Background: This study was designed to evaluate the matching percentage among findings of the ultrasound scans to the magnetic resonance imaging in women with fibroids in Bahrain.Methods: This descriptive, retrospective study was conducted from January 2016 to December 2018 including all the female patients referred from the Gynaecological Department to the Radiological Department in the Bahrain Defense Force Hospital for magnetic resonance imaging and ultrasound scan for fibroid evaluation.The data collected included their age, parity and nationality. The StatsDirect software was used to analyze the fibroids according to the site, size, type and number. Results: The average age of the recruited 205 female patients for the study was 43 years, with 81.5%as Bahraini citizens and a mean parity of 2.3. Ultrasound scan findings matched the Magnetic Resonance Imaging in the posterior fibroid sites in 52.5% of cases, followed by anterior (38.1%) and fundal (21.8%). For fibroid size, Ultrasound scans matched Magnetic Resonance Imaging findings in 83.3% for fibroids between >5 to ≤ 10 cm, in 78.3% for fibroids between >2 to ≤ 5cm, and in 36.8% for fibroids sized ≤ 2 cm. Fibroids more than 10cm in size were in agreement for 33.33% of fibroids. Submucosal fibroids matched in just 29.4% of cases, but for the subserosal fibroids, it was 44.8%. The matching score for more than 4 fibroids was 61.8%, followed by single fibroids (54.8%) and lastly for two fibroids (34.8%).Conclusions: Fibroid site and size had the highest matching rates amongst all the categories.