Dual-band refractive index sensor with cascaded asymmetric resonant compound grating based on bound states in the continuum

We propose a cascaded asymmetric resonant compound grating (ARCG) for high-performance dual-band refractive index sensing. The physical mechanism of the sensor is investigated using a combination of temporal coupled-mode theory (TCMT) and ARCG eigenfrequency information, which is verified by rigorous coupled-wave analysis (RCWA). The reflection spectra can be tailored by changing the key structural parameters. And by altering the grating strip spacing, a dual-band quasi-bound state in the continuum can be achieved. The simulation results show that the highest sensitivity of the dual-band sensor is 480.1 nm/RIU, and its figure of merit is 4.01 × 10⁵. The proposed ARCG has potential application prospects for high-performance integrated sensors.

Conformational entropy of hyaluronic acid contributes to taste enhancement

Natural taste/flavor enhancers are essential ingredients that could potentially address condiments overconsumption. For the first time, we report that hyaluronic acid (HA) could modulate taste perception, governed by the dynamic interactions among taste compounds, mucin, and HA. Various conformations of HA impact taste perception. The high molecular weight (Mw) of 1090 kDa HA inhibits the sense of taste due to its increased viscosity, which hinders the penetration of Na⁺ into the mucin layer. HA with low and medium Mw (100 kDa, 400 kDa) could enhance taste perception. Isothermal titration calorimetry analysis confirms the stronger binding between mucin and HA. The intensity of their interaction increases as the Mw of HA increases from 8 kDa to 400 kDa. Quartz crystal microbalance with dissipation characterization further indicates that the rigid conformation of 100 kDa HA facilitates the binding of Na⁺ with taste receptors, thereby enhancing taste perception. The flexible conformation of 400 kDa HA may conceal the taste receptor cells, reducing taste enhancement. Our work advances the understanding of conformational entropy of natural mucoadhesion and mucopenetration polymers, which lays the foundation for their potential use as taste enhancers.

[HSDL2 overexpression promotes rectal cancer progression by regulating cancer cell cycle and promoting cell proliferation]

OBJECTIVE

To analyze the expression of hydroxysteroid dehydrogenase like 2 (HSDL2) in rectal cancer tissues and the effect of changes in HSDL2 expression level on proliferation of rectal cancer cells.

METHODS

Clinical data and tissue samples of 90 patients with rectal cancer admitted to our hospital from January 2020 to June 2022 were collected from the prospective clinical database and biological specimen database. The expression level of HSDL2 in rectal cancer and adjacent tissues was detected by immunohistochemistry, and based on the median level of HSDL2 expression, the patients were divided into high expression group (n=45) and low expression group (n=45) for analysis the correlation between HSDL2 expression level and the clinicopathological parameters. GO and KEGG enrichment analyses were performed to explore the role of HSDL2 in rectal cancer progression. The effects of changes in HSDL2 expression levels on rectal cancer cell proliferation, cell cycle and protein expressions were investigated in SW480 cells with lentivirus-mediated HSDL2 silencing or HSDL2 overexpression using CCK-8 assay, flow cytometry and Western blotting.

RESULTS

The expressions of HSDL2 and Ki67 were significantly higher in rectal cancer tissues than in the adjacent tissues (P < 0.05). Spearman correlation analysis showed that the expression of HSDL2 protein was positively correlated with Ki67, CEA and CA19-9 expressions (P < 0.01). The rectal cancer patients with high HSDL2 expressions had significantly higher likelihood of having CEA ≥5 μg/L, CA19-9 ≥37 kU/L, T3-4 stage, and N2-3 stage than those with a low HSDL2 expression (P < 0.05). GO and KEGG analysis showed that HSDL2 was mainly enriched in DNA replication and cell cycle. In SW480 cells, HSDL2 overexpression significantly promoted cell proliferation, increased cell percentage in S phase, and enhanced the expression levels of CDK6 and cyclinD1 (P < 0.05), and HSDL2 silencing produced the opposite effects (P < 0.05).

CONCLUSION

The high expression of HSDL2 in rectal cancer participates in malignant progression of the tumor by promoting the proliferation and cell cycle progress of the cancer cells.

Case report: Fulminant type 1 diabetes following paucisymptomatic SARS-CoV-2 infection during late pregnancy

Background

Dysregulation of glucose metabolism has been linked to SARS-CoV-2 infection. In addition, the occurrence of new onset diabetes mellitus, including fulminant type 1 diabetes, has been reported after SARS-CoV-2 infection or vaccination.

Methods and results

A young Chinese woman in her last trimester of pregnancy presented with an abrupt progression of hyperglycemia and ketoacidosis, but with a near-normal glycohemoglobin level following paucisymptomatic SARS-CoV-2 infection. The low C peptide levels, both fasting and postprandial, reflected profound insulin deficiency in the setting of negative islet autoantibody testing, consistent with a diagnosis of fulminant type 1 diabetes. Ketoacidosis and hyperglycemia quickly improved following the introduction of insulin therapy, but not the β cell function. The patient received treatment with insulin pump therapy after being discharged, and the first follow-up revealed a well-controlled glucose profile.

Conclusions

New-onset FT1D can occur after SARS-CoV-2 infection. Our report raises awareness of this rare but serious situation, promoting early recognition and management of FT1D during the COVID-19 pandemic.

Calorie restriction ameliorates hyperglycemia, modulates the disordered gut microbiota, and mitigates metabolic endotoxemia and inflammation in type 2 diabetic rats

PURPOSE

The effects of calorie restriction (CR) on gut microbiota and the mechanism of CR ameliorating hyperglycemia in streptozotocin (STZ)-induced T2DM model rats were explored.

METHODS

High-fat diet and STZ injection were applied to induce T2DM model rats. Rats were divided into the following three groups: the control-diet ad libitum group, the T2DM model group fed with ad libitum diet, and the T2DM group fed with 30% restriction diet. 16S rRNA sequencing was used to determine the bacterial communities. Lipopolysaccharide (LPS)-binding protein (LBP), interleukin-6 (IL-6), and tumor necrosis factor-α (TNF-α) were measured.

RESULTS

Glucose tolerance and insulin sensitivity were improved by CR, as well as the levels of fasting and random plasma glucose. Besides, CR not only modulated the overall structure of gut microbiota but also had selective enrichment in anti-inflammatory bacteria such as Lachnospiraceae_NK4A136_group, Ruminococcaceae_9, Allobaculum, Alistipes, and Oscillibacter, and decreased pro-inflammatory pathogenic bacteria such as Bacteroides, Lachnoclostridium, and Bifidobacterium. Tax4Fun indicated that CR could regulate related functional pathways such as lipopolysaccharide biosynthesis, and the plasma levels of LBP, IL-6, and TNF-α were markedly reduced by CR, suggesting the mechanism of CR ameliorating hyperglycemia may associate with the modulation of disordered gut microbiota and the reduction of metabolic endotoxemia and inflammation.

CONCLUSION

CR could ameliorate hyperglycemia, the mechanism of which may associate with the alteration of the overall structure of gut microbiota, restoration of disordered microbiota function, and the downregulation of metabolic endotoxemia and inflammation in diabetic rats.

Periderm Fate during Palatogenesis: TGF-β and Periderm Dedifferentiation

Failure of palatogenesis results in cleft palate, one of the most common congenital disabilities in humans. During the final phases of palatogenesis, the protective function of the peridermal cell layer must be eliminated for the medial edge epithelia to adhere properly, which is a prerequisite for the successful fusion of the secondary palate. However, a deeper understanding of the role and fate of the periderm in palatal adherence and fusion has been hampered due to a lack of appropriate periderm-specific genetic tools to examine this cell type in vivo. Here we used the cytokeratin-6A (Krt-6a) locus to develop both constitutive (Krt6ai-Cre) and inducible (Krt6ai-CreERT2) periderm-specific Cre driver mouse lines. These novel lines allowed us to achieve both the spatial and temporal control needed to dissect the periderm fate on a cellular resolution during palatogenesis. Our studies suggest that, already before the opposing palatal shelves contact each other, at least some palatal periderm cells start to gradually lose their squamous periderm-like phenotype and dedifferentiate into cuboidal cells, reminiscent of the basal epithelial cells seen in the palatal midline seam. Moreover, we show that transforming growth factor-β (TGF-β) signaling plays a critical periderm-specific role in palatogenesis. Thirty-three percent of embryos lacking a gene encoding the TGF-β type I receptor (Tgfbr1) in the periderm display a complete cleft of the secondary palate. Our subsequent experiments demonstrated that Tgfbr1-deficient periderm fails to undergo appropriate dedifferentiation. These studies define the periderm cell fate during palatogenesis and reveal a novel, critical role for TGF-β signaling in periderm dedifferentiation, which is a prerequisite for appropriate palatal epithelial adhesion and fusion.

Aneurysmal bone cyst of the temporal bone presenting with reversible vestibular impairment

BACKGROUND

Aneurysmal bone cysts are expansile benign lesions associated with compressive destruction and obscure pathogenesis. The most common sites of temporal bone involvement are the petrous apex, squamous portions and mastoid.

CASE REPORT

This paper reports a right temporal aneurysmal bone cyst in a 51-year-old man who presented clinically with facial palsy, and hearing loss and impaired vestibular function. Magnetic resonance imaging and computed tomography findings were consistent with a diagnosis of aneurysmal bone cyst. Inter-operative findings showed that the lesion had caused compressive damage to the internal auditory canal. Following surgical excision, the patient experienced vertigo, indicating recovery of vestibular function. Follow-up imaging revealed complete resection without clinical recurrence.

CONCLUSION

To our knowledge, this is the first report of aneurysmal bone cyst invasion of the inner auditory canal. Our clinical experience indicates that vestibular nerve damage recovery is relatively uncommon. This case report will hopefully inform future studies.

Research advances in ZnO nanomaterials-based UV photode tectors: a review

Ultraviolet photodetectors (UV PDs) have always been the research focus of semiconductor optoelectronic devices due to their wide application fields and diverse compositions. As one of the best-known n-type metal oxides in third-generation semiconductor electronic devices, ZnO nanostructures and their assembly with other materials have received extensive research. In this paper, the research progress of different types of ZnO UV PDs is reviewed, and the effects of different nanostructures on ZnO UV PDs are summarized in detail. In addition, physical effects such as piezoelectric photoelectric effect, pyroelectric effect, and three ways of heterojunction, noble metal local surface plasmon resonance enhancement and formation of ternary metal oxides on the performance of ZnO UV PDs were also investigated. The applications of these PDs in UV sensing, wearable devices, and optical communication are displayed. Finally, the possible opportunities and challenges for the future development of ZnO UV PDs are prospected.

[CD5L is elevated in the serum of patients with candidemia and promotes disease progression in mouse models]

OBJECTIVE

To investigate the changes of CD5L levels in patients with candidemia and explore the role of CD5L in progression of candidemia.

METHODS

Twenty healthy control individuals, 27 patients with bacteremia and 35 patients with candidemia were examined for serum CD5L levels using ELISA, and the correlations of CD5L level with other serological indicators were analyzed. A C57BL/6 mouse model of candidemia induced by intravenous injection of Candida albicans were treated with intraperitoneal injection of recombinant CD5L protein, and renal histopathological and serological changes were analyzed to assess renal injures. The effects of CD5L treatment on general condition, fungal burden, of survival of the mice were observed, and the changes in serum IL-6 and IL-8 levels of the mice were detected using ELISA.

RESULTS

CD5L levels were significantly elevated in patients with candidemia and positively correlated with WBC, BDG, Scr and PCT levels. The mouse model of candidemia also showed significantly increased serum and renal CD5L levels, and CD5L treatment significantly increased fungal burden in the renal tissue, elevated IL-6 and IL-8 levels in the serum and kidney, aggravated renal tissue damage, and reduced survival rate of candidemia mice.

CONCLUSION

Serum CD5L levels are increased in patients with candidemia, and treatment with CD5L aggravates candidemia in mouse models.

Recognition and localization of asymmetric spectra in FBG sensing networks

We propose a deep learning demodulation method based on a long short-term memory (LSTM) neural network for fiber Bragg grating (FBG) sensing networks. Interestingly, we find that both low demodulation error and distorted spectrum recognition are realized using the proposed LSTM-based method. Compared with conventional demodulation methods, including Gaussian-fitting, convolutional neural network, and the gated recurrent unit, the proposed method improves the demodulation accuracy being close to 1 pm and achieves a demodulation time of 0.1s for 128-FBG sensors. Furthermore, our approach can realize 100% accuracy of distorted spectra recognition and complete the location of spectra with spectrally encoded FBG sensors.

Promotional Effect and Molecular Mechanism of Synthesized Zinc Oxide Nanocrystal on Zirconia Abutment Surface for Soft Tissue Sealing

Studies have confirmed that tooth loss is closely related to systemic diseases, such as obesity, diabetes, cardiovascular diseases, some types of tumors, and Alzheimer's disease. Among many methods for tooth restoration, implant restoration is the most commonly used method. After implantation, long-term stability of implants requires not only good bone bonding but also good soft tissue sealing between implants and surrounding soft tissues. The zirconia abutment is used in clinical implant restoration treatment, but due to the strong biological inertia of zirconia, it is difficult to form stable chemical or biological bonds with surrounding tissues. In this study, we investigated synthesized zinc oxide (ZnO) nanocrystal on the zirconia abutment surface by the hydrothermal method to make it more beneficial for soft tissue early sealing and the molecular mechanism. In vitro experiments found that different hydrothermal treatment temperatures affect the formation of ZnO crystals. The crystal diameter of ZnO changes from micron to nanometer at different temperatures, and the crystal morphology also changes. In vitro, scanning electron microscopy, energy dispersive spectrometry, and real-time polymerase chain reaction results show that ZnO nanocrystal can promote the attachment and proliferation of oral epithelial cells on the surface of zirconia by promoting the binding of laminin 332 and integrin β4, regulating the PI3K/AKT pathway. In vivo, ZnO nanocrystal ultimately promotes the formation of soft tissue seals. Collectively, ZnO nanocrystal can be synthesized on a zirconia surface by hydrothermal treatment. It can help to form a seal between the implant abutment and surrounding soft tissue. This method is beneficial to the long-term stability of the implant and also can be applied to other medical fields.

Evidence of Antineutrinos from Distant Reactors Using Pure Water at SNO

The SNO+ Collaboration reports the first evidence of reactor antineutrinos in a Cherenkov detector. The nearest nuclear reactors are located 240 km away in Ontario, Canada. This analysis uses events with energies lower than in any previous analysis with a large water Cherenkov detector. Two analytical methods are used to distinguish reactor antineutrinos from background events in 190 days of data and yield consistent evidence for antineutrinos with a combined significance of 3.5σ.

[Interventional treatment and early-stage follow-up of pulmonary atresia with intact ventricular septum]

Objective: To explore the application value of percutaneous peripheral interventional therapy in pulmonary atresia with intact ventricular septal (PA-IVS). Methods: Retrospective case summary. The data was collected from 25 children who were hospitalized at the Children's Hospital,Zhejiang University School of Medicine from August 2019 to August 2022, had been diagnosed with PA-IVS by echocardiography, and underwent interventional treatment. The sex, age, weight, operation time, radiation exposure time, and radiation dose of the patients were collected. The patients were divided into the arterial duct stenting group and the non-stenting group. Preoperative tricuspid annular diameters and Z scores, right ventricular length diameters, and right ventricular/left ventricular length-diameter ratios were compared by paired t-tests. Right ventricular systolic pressure difference, oxygen saturation, lactic acid before and after the surgery were compared for 24 children who received percutaneous balloon pulmonary valvuloplasty. Right ventricular improvement in 25 children after operation was analyzed. The correlation between postoperative oxygen saturation and postoperative right ventricular systolic blood pressure difference, the degree of pulmonary valve opening and the Z value of tricuspid valve ring in the non-stenting group were analyzed. Results: A total of 25 patients with PA-IVS were enrolled in the study, of whom 19 were males and 6 females, with an age at surgery of 12 (6, 28) days and a weight of (3.7±0.5) kg. One of them underwent only stenting of the arterial duct; 20 children underwent only percutaneous pulmonary valve perforation and balloon angioplasty; 4 children underwent both procedures. The Z-value of the tricuspid ring was -1.5±1.2 in the group with arterial duct stenting, and -0.1±0.4 in the group without stenting (t=2.77, P=0.010). The tricuspid regurgitant flow rate 1 month after surgery was significantly lower than the preoperative ((3.4±0.6) vs. (4.8±0.9) m/s, t=6.62,P<0.001). In the 24 children with percutaneous pulmonary valve perforation and balloon angioplasty, the preoperative right ventricular systolic blood pressure was (110±32) mmHg, and the postoperative systolic blood pressure was (52±19) mmHg (1 mmHg=0.133 kPa) (F=59.55, P<0.001). The factors that may affect postoperative oxygen saturation in 20 cases of non-stenting group were analyzed. The results suggested that the pre and post-operative right ventricular systolic blood pressure differences (r=-0.11, P=0.649), and the pulmonary valve orifice opening (r=-0.31, P=0.201) and tricuspid annulus Z value (r=-0.18, P=0.452) at 1 month after the operation were not significantly correlated with the postoperative oxygen saturation. Conclusions: Interventional therapy can be used as the first choice for one-stage operation of PA-IVS. Percutaneous pulmonary valve perforation and balloon angioplasty are more suitable for children with well-developed right ventricles, tricuspid annulus, and pulmonary arteries. While the smaller the tricuspid annulus, the more dependent it is on the ductus arteriosus and thus patients are more suitable for arterial duct stenting.

CircNDST1 promotes papillary thyroid cancer progression via its interaction with CSNK2A1 to activate the PI3K-Akt pathway and epithelial-mesenchymal transition

BACKGROUND

Multiple studies have established a strong relationship between circRNA and cancer progression. Cervical lymph node metastasis is a key factor influencing the surgical approach and distant metastasis of papillary thyroid cancer (PTC). However, the role of circNDST1 in PTC has not been investigated. Our research focused on revealing the function and mechanism of action of circNDST1 in PTC.

METHODS

High-throughput sequencing and qPCR were used to assess the expression of circRNA in PTC tissues with extensive cervical lymph node metastasis and circNDST1 in cell lines, respectively. The proliferative effects of circNDST1 in vitro and in vivo were analyzed using CCK8, clone formation assay, EdU, and nude mouse tumorigenesis assay. The transwell scratch assay was employed in the scrutiny of the effect of circNDST1 on the migration and invasion abilities of thyroid cancer cells, while circNDST1's influence on the PI3K-Akt pathway and the Epithelial-Mesenchymal Transition (EMT) key protein expression was evaluated utilizing RNA sequencing and western blot. RNA pull-down and RIP were used to examine the binding of circNDST1 to CSNK2A1.

RESULTS

CircNDST1 was highly expressed in PTC cell lines, but knocking it down inhibited the proliferation, migration, and invasive abilities of TPC1 and KTC1 cell lines. CircNDST1 bonded with CSNK2A1 and promoted the interaction between CSNK2A1 and Akt, leading to the activation of the PI3K-Akt pathway and EMT.

CONCLUSION

CircNDST1's high expression boosted thyroid cancer progression through the activation of the PI3K-Akt pathway and EMT in a CSNK2A1-dependent manner.

Microarray profiling identifies hsa_circ_0082003 as a novel tumor promoter for papillary thyroid carcinoma

BACKGROUND

Circular RNAs (circRNAs) are non-coding RNAs that have essential regulatory roles in the development of various tumors. This study explored whether circRNAs are involved in the progression of papillary thyroid carcinoma (PTC).

METHODS

Differentially expressed circRNAs (DECs) in four pairs of PTC and matched normal thyroid tissues were screened using a circRNA microarray. The potential functions of dysregulated circRNAs were predicted by bioinformatic analyses. Reverse transcription quantitative polymerase chain reaction (RT-qPCR) was used to determine hsa_circ_0082003 expression in 80 pairs of PTC and matched normal thyroid tissues. Cell counting kit-8, colony formation, wound healing, and Transwell assays were performed to evaluate the biological functions of hsa_circ_0082003 in PTC cells. The role of hsa_circ_0082003 in PTC tumorigenesis in vivo was validated in nude mice.

RESULTS

In total, 3150 DECs (2317 upregulated and 833 downregulated) were identified. Pathway enrichment analyses indicated that the dysregulated circRNAs may play roles in PTC development. RT-qPCR validation demonstrated that hsa_circ_0082003 expression was significantly increased in PTC tissues and correlated with poor clinicopathological parameters. Receiver operating characteristic curve analysis showed that hsa_circ_0082003 had good performance for diagnosing PTC and judging whether it was accompanied by lymph node metastasis. Knockdown of hsa_circ_0082003 inhibited PTC cell proliferation, migration, and invasion. Tumor formation assays in vivo showed that downregulation of hsa_circ_0082003 significantly suppressed the growth of PTC.

CONCLUSION

Hsa_circ_0082003 may serve as a novel diagnostic biomarker and potential therapeutic target for PTC.

Efficacy and safety of apatinib and transcatheter arterial chemoembolization as second-line therapy for advanced hepatocellular carcinoma: A retrospective cohort study

Background

Transarterial chemoembolization (TACE) and tyrosine kinase inhibitors (TKIs) have demonstrated clinical efficacy against hepatocellular carcinoma (HCC) as first-line therapy. However, there is limited evidence on the efficacy and safety of apatinib plus TACE as second-line treatment for advanced HCC.

Aim

To evaluate the efficacy and safety of apatinib in combination with TACE for advanced HCC patients with disease progression or intolerant to first-line treatment.

Patients and Methods

Between May 2019 and January 2022, 72 advanced HCC patients received apatinib plus TACE as second-line treatment. Clinical parameters, efficacy, and safety were assessed. The primary endpoint was progression-free survival (PFS) and the secondary the objective response rate (ORR) and disease control rate (DCR).

Results

The median follow-up duration was 14.7 months (range, 4.5-26.0). The median PFS from the beginning of treatment was 7.1 months (1.0-15.2) and its 95% CI was 6.6-8.2 according to Kaplan-Meier analysis. The ORR and DCR were 34.7% (95% CI: 23.9%-46.9%) and 48.6% (95% CI: 36.7%-60.7%), respectively. By the cutoff date, 33 patients (45.8%) had died and 39 (54.2%) were undergoing survival follow-up. The estimated median overall survival (mOS) was 22.3 months (95% CI: 20.6-24.0) by Kaplan-Meier analysis. The most common apatinib-related adverse events of any grade were hypertension [35 (48.6%)], appetite loss [30 (41.6%)], and hand-foot syndrome [21 (29.2%)].

Conclusion

The combination of apatinib and TACE as second-line therapy demonstrated promising clinical efficacy and tolerable adverse effects for advanced HCC patients.

A deep neural network for parametric image reconstruction on a large axial field-of-view PET

PURPOSE

The PET scanners with long axial field of view (AFOV) having ~ 20 times higher sensitivity than conventional scanners provide new opportunities for enhanced parametric imaging but suffer from the dramatically increased volume and complexity of dynamic data. This study reconstructed a high-quality direct Patlak Ki image from five-frame sinograms without input function by a deep learning framework based on DeepPET to explore the potential of artificial intelligence reducing the acquisition time and the dependence of input function in parametric imaging.

METHODS

This study was implemented on a large AFOV PET/CT scanner (Biograph Vision Quadra) and twenty patients were recruited with ¹⁸F-fluorodeoxyglucose (¹⁸F-FDG) dynamic scans. During training and testing of the proposed deep learning framework, the last five-frame (25 min, 40-65 min post-injection) sinograms were set as input and the reconstructed Patlak Ki images by a nested EM algorithm on the vendor were set as ground truth. To evaluate the image quality of predicted Ki images, mean square error (MSE), peak signal-to-noise ratio (PSNR), and structural similarity index measure (SSIM) were calculated. Meanwhile, a linear regression process was applied between predicted and true Ki means on avid malignant lesions and tumor volume of interests (VOIs).

RESULTS

In the testing phase, the proposed method achieved excellent MSE of less than 0.03%, high SSIM, and PSNR of ~ 0.98 and ~ 38 dB, respectively. Moreover, there was a high correlation (DeepPET: [Formula: see text]= 0.73, self-attention DeepPET: [Formula: see text]=0.82) between predicted Ki and traditionally reconstructed Patlak Ki means over eleven lesions.

CONCLUSIONS

The results show that the deep learning-based method produced high-quality parametric images from small frames of projection data without input function. It has much potential to address the dilemma of the long scan time and dependency on input function that still hamper the clinical translation of dynamic PET.

APASL-ACLF Research Consortium-Artificial Intelligence (AARC-AI) model precisely predicts outcomes in acute-on-chronic liver failure patients

BACKGROUND AND AIMS

We hypothesized that artificial intelligence (AI) models are more precise than standard models for predicting outcomes in acute-on-chronic liver failure (ACLF).

METHODS

We recruited ACLF patients between 2009 and 2020 from APASL-ACLF Research Consortium (AARC). Their clinical data, investigations and organ involvement were serially noted for 90-days and utilized for AI modelling. Data were split randomly into train and validation sets. Multiple AI models, MELD and AARC-Model, were created/optimized on train set. Outcome prediction abilities were evaluated on validation sets through area under the curve (AUC), accuracy, sensitivity, specificity and class precision.

RESULTS

Among 2481 ACLF patients, 1501 in train set and 980 in validation set, the extreme gradient boost-cross-validated model (XGB-CV) demonstrated the highest AUC in train (0.999), validation (0.907) and overall sets (0.976) for predicting 30-day outcomes. The AUC and accuracy of the XGB-CV model (%Δ) were 7.0% and 6.9% higher than the standard day-7 AARC model (p < .001) and 12.8% and 10.6% higher than the day 7 MELD for 30-day predictions in validation set (p < .001). The XGB model had the highest AUC for 7- and 90-day predictions as well (p < .001). Day-7 creatinine, international normalized ratio (INR), circulatory failure, leucocyte count and day-4 sepsis were top features determining the 30-day outcomes. A simple decision tree incorporating creatinine, INR and circulatory failure was able to classify patients into high (~90%), intermediate (~60%) and low risk (~20%) of mortality. A web-based AARC-AI model was developed and validated twice with optimal performance for 30-day predictions.

CONCLUSIONS

The performance of the AARC-AI model exceeds the standard models for outcome predictions in ACLF. An AI-based decision tree can reliably undertake severity-based stratification of patients for timely interventions.