Explainable cancer factors discovery: Shapley additive explanation for machine learning models demonstrates the best practices in the case of pancreatic cancer

Pancreatic cancer is one of digestive tract cancers with high mortality rate. Despite the wide range of available treatments and improvements in surgery, chemotherapy, and radiation therapy, the five-year prognosis for individuals diagnosed pancreatic cancer remains poor. There is still research to be done to see if immunotherapy may be used to treat pancreatic cancer. The goals of our research were to comprehend the tumor microenvironment of pancreatic cancer, found a useful biomarker to assess the prognosis of patients, and investigated its biological relevance. In this paper, machine learning methods such as random forest were fused with weighted gene co-expression networks for screening hub immune-related genes (hub-IRGs). LASSO regression model was used to further work. Thus, we got eight hub-IRGs. Based on hub-IRGs, we created a prognosis risk prediction model for PAAD that can stratify accurately and produce a prognostic risk score (IRG_Score) for each patient. In the raw data set and the validation data set, the five-year area under the curve (AUC) for this model was 0.9 and 0.7, respectively. And shapley additive explanation (SHAP) portrayed the importance of prognostic risk prediction influencing factors from a machine learning perspective to obtain the most influential certain gene (or clinical factor). The five most important factors were TRIM67, CORT, PSPN, SCAMP5, RFXAP, all of which are genes. In summary, the eight hub-IRGs had accurate risk prediction performance and biological significance, which was validated in other cancers. The result of SHAP helped to understand the molecular mechanism of pancreatic cancer.

Identification and validation of an explainable prediction model of acute kidney injury with prognostic implications in critically ill children: a prospective multicenter cohort study

Background

Acute kidney injury (AKI) is a common and serious organ dysfunction in critically ill children. Early identification and prediction of AKI are of great significance. However, current AKI criteria are insufficiently sensitive and specific, and AKI heterogeneity limits the clinical value of AKI biomarkers. This study aimed to establish and validate an explainable prediction model based on the machine learning (ML) approach for AKI, and assess its prognostic implications in children admitted to the pediatric intensive care unit (PICU).

Methods

This multicenter prospective study in China was conducted on critically ill children for the derivation and validation of the prediction model. The derivation cohort, consisting of 957 children admitted to four independent PICUs from September 2020 to January 2021, was separated for training and internal validation, and an external data set of 866 children admitted from February 2021 to February 2022 was employed for external validation. AKI was defined based on serum creatinine and urine output using the Kidney Disease: Improving Global Outcome (KDIGO) criteria. With 33 medical characteristics easily obtained or evaluated during the first 24 h after PICU admission, 11 ML algorithms were used to construct prediction models. Several evaluation indexes, including the area under the receiver-operating-characteristic curve (AUC), were used to compare the predictive performance. The SHapley Additive exPlanation method was used to rank the feature importance and explain the final model. A probability threshold for the final model was identified for AKI prediction and subgrouping. Clinical outcomes were evaluated in various subgroups determined by a combination of the final model and KDIGO criteria.

Findings

The random forest (RF) model performed best in discriminative ability among the 11 ML models. After reducing features according to feature importance rank, an explainable final RF model was established with 8 features. The final model could accurately predict AKI in both internal (AUC = 0.929) and external (AUC = 0.910) validations, and has been translated into a convenient tool to facilitate its utility in clinical settings. Critically ill children with a probability exceeding or equal to the threshold in the final model had a higher risk of death and multiple organ dysfunctions, regardless of whether they met the KDIGO criteria for AKI.

Interpretation

Our explainable ML model was not only successfully developed to accurately predict AKI but was also highly relevant to adverse outcomes in individual children at an early stage of PICU admission, and it mitigated the concern of the "black-box" issue with an undirect interpretation of the ML technique.

Funding

The National Natural Science Foundation of China, Jiangsu Province Science and Technology Support Program, Key talent of women's and children's health of Jiangsu Province, and Postgraduate Research & Practice Innovation Program of Jiangsu Province.

The Relationship between Acid Production and the Microbial Community of Newly Produced Coal Gangue in the Early Oxidation Stage

Coal gangue is a solid waste formed during coal production, and the acid mine drainage it generates during open-pit storage severely pollutes the ecological environment of mining areas. Microorganisms play a crucial catalytic role in acidification, and their species and gene functions change during the oxidation process of coal gangue. In this study, the changes in microbial community structure were investigated during the initial acidification process for newly produced gangue exposed to moisture by monitoring the changes in pH, EC, sulfate ion concentration, and the iron oxidation rate of gangue leaching solutions. Moreover, the composition and functional abundance of microbial communities on the surface of the gangue were analyzed with rainfall simulation experiments and 16S rRNA sequencing. The study yielded the following findings: (1) The critical period for newly produced gangue oxidation spanned from 0~15 d after its exposure to water; the pH of leaching solutions decreased from 4.65 to 4.09 during this time, and the concentration and oxidation rate of iron in the leaching solutions remained at low levels, indicating that iron oxidation was not the main driver for acidification during this stage. (2) When the gangue was kept dry, Burkholderia spp. dominated the gangue microbial community. When the gangue was exposed to moisture, the rate of acidification accelerated, and Pseudomonas replaced Burkholderia as the dominant genus in the community. (3) In terms of gene function, the microbial community of the acidified gangue had stronger nitrogen cycling functions, and an increase in the abundance of microorganisms related to the sulfur cycle occurred after day 15 of the experiment. The microbial community in the acidified gangue had more stress resistance than the community of the newly formed gangue, but its potential to decompose environmental pollutants decreased.

[Research on indicators of ideological and political resource database construction for curriculum of "Epidemiology"]

Objective: To construct indicators of the ideological and political resource database construction for the curriculum of "Epidemiology". Methods: Two rounds of expert consultation were conducted in 15 experts from 4 universities and 1 textbook publishing house using the Delphi method, and the importance and feasibility scores of the indicators were calculated with the degree of concentration and coordination of experts' opinions. Results: In the two rounds of consultation, the experts' positive coefficient of the two questionnaires were both 100.00% (15/15), the authoritative coefficients of experts were both 0.83, and the Kendall's W was 0.27 (P<0.05) and 0.33 (P<0.05), respectively. Consensus was reached on 4 primary indicators and 31 secondary indicators. Conclusion: The process of this study is scientific, and the indicators for the construction of ideological and political resource database for the curriculum of "Epidemiology" are authoritative, which can promote the establishment of ideological and political resource database for the curriculum of "Epidemiology".

[A descriptive analysis of hyperlipidemia in adult twins in China]

Objective: To describe the distribution characteristics of hyperlipidemia in adult twins in the Chinese National Twin Registry (CNTR) and explore the effect of genetic and environmental factors on hyperlipidemia. Methods: Twins recruited from the CNTR in 11 project areas across China were included in the study. A total of 69 130 (34 565 pairs) of adult twins with complete information on hyperlipidemia were selected for analysis. The random effect model was used to characterize the population and regional distribution of hyperlipidemia among twins. The concordance rates of hyperlipidemia were calculated in monozygotic twins (MZ) and dizygotic twins (DZ), respectively, to estimate the heritability. Results: The age of all participants was (34.2±12.4) years. This study's prevalence of hyperlipidemia was 1.3% (895/69 130). Twin pairs who were men, older, living in urban areas, married,had junior college degree or above, overweight, obese, insufficient physical activity, current smokers, ex-smokers, current drinkers, and ex-drinkers had a higher prevalence of hyperlipidemia (P<0.05). In within-pair analysis, the concordance rate of hyperlipidemia was 29.1% (118/405) in MZ and 18.1% (57/315) in DZ, and the difference was statistically significant (P<0.05). Stratified by gender, age, and region, the concordance rate of hyperlipidemia in MZ was still higher than that in DZ. Further, in within-same-sex twin pair analyses, the heritability of hyperlipidemia was 13.04% (95%CI: 2.61%-23.47%) in the northern group and 18.59% (95%CI: 4.43%-32.74%) in the female group, respectively. Conclusions: Adult twins were included in this study and were found to have a lower prevalence of hyperlipidemia than in the general population study, with population and regional differences. Genetic factors influence hyperlipidemia, but the genetic effect may vary with gender and area.

[Clinical research progress and implications of therapeutic vaccines for cervical cancer and precancerous lesions: a qualitative systematic review]

Objective: To systematically summarize and analyze the clinical research progress of therapeutic vaccines for cervical cancer or precancerous lesions. Methods: English databases (PubMed, Embase, Web of Science, Cochrane library, Proquest, and ClinicalTrails.gov) and Chinese databases (SinoMed, CNKI, WanFang, and VIP Database) were systematically searched to collect literature on therapeutic vaccines for cervical cancer or precancerous lesions from inception to February 18, 2021. After screening, we evaluated the risk of bias of included studies, and combed the basic information of the literature, research designs, information of vaccines, study patients, outcome indicators and so on, qualitatively summarized the clinical research progress. Results: A total of 71 studies were included in this systematic review, including 14 random controlled trials, 15 quasi-random controlled trials, 4 cohort studies, 1 case-control study, 34 case series studies and 3 case reports. The study patients included women aged 15~79 with cervical cancer or precancerous lesions in 18 countries from 1989 to 2021. On the one hand, there were 40 studies on therapeutic vaccines for cervical precancerous lesions (22 867 participants), involving 21 kinds of vaccines in 6 categories. Results showed 3 marketed vaccines (Cervarix, Gardasil, Gardasil 9) as adjuvant immunotherapies were significant effective in preventing the recurrence of precancerous lesions compared with the conization only. In addition, MVA E2 vaccine had been in phase Ⅲ clinical trials as a specific therapeutic vaccine, with relative literature showing it could eliminate most high-grade precancerous lesions. Therapeutic vaccines for precancerous lesions all showed good safety. On the other hand, there were 31 studies on therapeutic vaccines for cervical cancer (781 participants), involving 19 kinds of vaccines in 7categories, with none had been marketed. 25 studies were with no control group, showing the vaccines could effectively eliminate solid tumors, prevent recurrence, and prolong the median survival time. However, the vaccines effectiveness couldn't be statistically calculated due to the lack of a control group. As for the safety of therapeutic vaccines for cervical cancer, 9 studies showed that patients experienced serious adverse events after treatments, where 7 studies reported that serious adverse events occurred in patients couldn't be ruled out as the results of therapeutic vaccines. Conclusions: The literature review shows that the literature evidence for the therapeutic vaccines for cervical precancerous lesions is relatively mature compared with the therapeutic vaccines for cervical cancer. The four kinds of vaccines on the market are all therapeutic vaccines for precancerous lesions, but they are generally used as vaginal infection treatments or adjuvant immunotherapies for cervical precancerous lesions, not used for the specific treatments of cervical precancerous lesions. Other specific therapeutic vaccines are in the early stage of clinical trials, mainly phase Ⅰ/Ⅱ clinical trials with small sample size. The effectiveness and safety data are limited, and further research is still needed.

High accuracy measurement of the ground-state hyperfine splitting in a 113Cd+ microwave clock: erratum

We present an erratum to our Letter [Opt. Lett.40, 4249 (2015)OPLEDP0146-959210.1364/OL.40.004249]. This erratum corrects the nuclear Lande factor gI in Eq. (2). After correcting the error, the final ground-state hyperfine splitting frequency of the ¹¹³Cd⁺ ion is determined to be 15199862855.0287(10) Hz.

High accuracy measurement of the ground-state hyperfine splitting in a ¹¹³Cd⁺ microwave clock

A microwave frequency standard based on laser-cooled (113)Cd(+) ions has been developed in recent years, and the short-term frequency instability is measured to be 6.1×10(-13)/√τ. By comparing the Cd(+) clock to a superior frequency reference, the ground-state hyperfine splitting of (113)Cd(+) is measured precisely to be 15199862855.0192(10) Hz with a fractional precision of 6.6×10(-14). This result is consistent with previous results, and the measurement precision is improved by nearly one order more than the best result reported before.

Note: A novel design of a microwave feed for a microwave frequency standard with a linear ion trap

Linear ion traps are important tools in many applications, particularly in mass spectrum analyzers and frequency standards. Here a novel design of a microwave feed integrated into one electrode of a linear quadrupole ion trap is demonstrated for the application of a microwave frequency standard based on cadmium ions. The mechanical structure of the microwave feed is compact and easy to build. The ion trap integrated with this microwave feed is successfully applied to measure the hyperfine splitting of the ground state of (113)Cd(+), thus demonstrating the practicality and reliability of the microwave feed.

Laser frequency stability transfer using a fiber-based interferometer

We have demonstrated a method for laser frequency stability transfer using a fiber-based Young's interferometer. An 858 nm external cavity diode laser is stabilized to within 1e-8 from 10 s to 4000 s, referenced to a Rubidium stabilized 780 nm DBR diode laser using the interferometer as a frequency-stability-transferring link. The system is simple to build and can link any two laser wavelengths within the fiber operating range.

High-accuracy measurement of the 113Cd+ ground-state hyperfine splitting at the milli-Hertz level

We have developed a microwave frequency standard based on the 15.2 GHz ground-stated hyperfine transition of (113)C(+) ions. Using a laser-cooled ion cloud trapped in a linear quadrupole Paul trap, the clock transition frequency is measured to be 15 199 862 855.0125(87) Hz, with an accuracy at the 10(-13) level. The main errors are from the microwave frequency reference used in the experiment. The precision is improved by nearly two orders of magnitude than that reported before.

Genome-wide linkage scan of 98 bipolar pedigrees and analysis of clinical covariates

Despite compelling evidence that genetic factors contribute to bipolar disorder (BP), attempts to identify susceptibility genes have met with limited success. This may be due to the genetic heterogeneity of the disorder. We sought to identify susceptibility loci for BP in a genome-wide linkage scan with and without clinical covariates that might reflect the underlying heterogeneity of the disorder. We genotyped 428 subjects in 98 BP families at the Center for Inherited Disease Research with 402 microsatellite markers. We first carried out a non-parametric linkage analysis with MERLIN, and then reanalyzed the data with LODPAL to incorporate clinical covariates for age at onset (AAO), psychosis and comorbid anxiety. We sought to further examine the top findings in the covariate analysis in an independent sample of 64 previously collected BP families. In the non-parametric linkage analysis, three loci were nominally significant under a narrow diagnostic model and seven other loci were nominally significant under a broader model. The top findings were on chromosomes 2q24 and 3q28. The covariate analyses yielded additional evidence for linkage on 3q28 with AAO in the primary and independent samples. Although none of the linked loci were genome-wide significant, their congruence with prior results and, for the covariate analyses, their identification in two separate samples increases the likelihood that they are true positives and deserve further investigation. These findings further demonstrate the value of considering clinical features that may reflect the underlying heterogeneity of disease in order to facilitate gene mapping.

Identification of two repressor elements in the mouse alpha 2(I) collagen promoter

We recently identified three areas of Sp1 binding located between -568 and -453 of the 5' flanking region of the murine alpha2(I) collagen promoter which are necessary for optimal activity. We now identify two additional regions of Sp1 binding located at -371 to -351 (region 4) and at -690 to -613 (region 5), which when mutated increased promoter activity in transfected rat hepatic stellate cells indicating they contain negative regulatory elements. AP-2 bound to region 4 while YY1 bound most strongly to region 5. AP-2 decreased Sp1 binding to region 4 and had a dual effect on Sp1 binding to region 5 decreasing and increasing Sp1 binding at low and high concentrations of AP-2, respectively. YY1 enhanced Sp1 binding to both regions. AP-2 inhibited or enhanced the stimulatory effect of a transfected Sp1 expression vector on the alpha2(I) collagen promoter in Drosophila cells at low or high AP-2 expression, respectively. YY1 enhanced or inhibited the activation of the promoter by low or high Sp1 expression, respectively. This study identifies two negative regulatory elements in the murine alpha2(I) collagen promoter and shows that AP-2 and YY1 interact with Sp1 at these sites and can inhibit the activating action of Sp1.

Effect of acetaldehyde on Sp1 binding and activation of the mouse alpha 2(I) collagen promoter

Acetaldehyde activates the mouse alpha 2(I) collagen promoter and this effect is mediated in part by increased binding of nuclear factor I (NF-I). Additional mechanisms may exist since deletions in the promoter upstream to the NF-I binding site prevented enhancement by acetaldehyde. Three adjacent areas of binding by nuclear proteins from activated hepatic stellate cells were identified at -568 to -554 (region 1), -542 to -518 (region 2), and -473 to -453 (region 3) of the promoter using DNase I protection analyses. Multiple DNA-protein complexes were formed in electrophoretic mobility shift assays with oligonucleotide probes specifying the three regions. Sp1 and NF-1 bound to all three regions, while Sp3 bound to region 2. Acetaldehyde decreased nuclear protein binding to all three regions. Mutations of regions 1, 2, and 3 reduced basal activity of the promoter and inhibited acetaldehyde stimulation in transfected stellate cells. Acetaldehyde inhibited the stimulatory effect of the Sp1 vector pPacSp1 on the promoter in transfected Drosophila cells. In conclusion, three regions of Sp1 binding were identified and are required for optimal activity of the alpha 2(I) collagen promoter. Sp1 is required for basal activity of the alpha 2(I) collagen promoter; however, the enhancing effect of acetaldehyde on the promoter is not mediated by Sp1.

Contributions of K+, Na+, and Cl- to the membrane potential of intact hamster vascular endothelial cells

The transmembrane potential (Vm) of vascular endothelial cells (EC) is an important property that may be involved in intra- and intercellular signal transduction for various vascular functions. In this study, Vm of intact aortic and vena caval EC from hamsters were measured using conventional microelectrodes. Vascular strips with the luminal surface upwards were suffused in a tissue chamber with Krebs solution in physiological conditions. The resting Vm of aortic and vena caval EC was found to be -40 +/- 1 mV (n = 55) and -43 +/- 1 mV (n = 15), respectively. The Vm recordings were confirmed to have originated from EC by scanning and transmission electron microscopy combined with the comparison of electrical recordings between normal and endothelium-denuded aortic strips. The input resistance varied from 10-240 M omega, which implied the presence of electrical coupling between vascular EC. Elevating the K+ level in the suffusate from 4.7 mM to 50 and 100 mM depolarized aortic EC by 19% and 29% and vena caval EC by 18% and 29%, respectively. These low percentages indicated a relatively small contribution of [K+] to the resting Vm of vascular EC. A positive correlation (r > 0.69) between the resting Vm and the magnitude of depolarization by the high [K+]o may be related to the involvement of voltage-dependent K+ channels. The hyperpolarization caused by lowering both [Na+]o and [Cl-]o suggested the disengagement of some electrogenic transport systems in the membrane, such as a Na(+)-K(+)-Cl- cotransporter. The transference number (t(ion)), as an index of membrane conductance for specific ions, was calculated for K+ (15-20%), Na+ (16%), and Cl- (9-15%), demonstrating that both Na+ and Cl- as well as K+ contribute to the overall resting Vm. Our study documented some basic electrophysiology of the vascular EC when both structural and functional properties of the cell were maintained, thus furthering the understanding of the essential role of endothelial cells in mediating vascular functions.