SEARCH · PubMed Health

Results for “predictive modelling”

Explore indexed PubMed citations for clinical trials, systematic reviews and public health research. Read source abstracts and follow each citation to its original PubMed record.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 703 records · Page 39Linked to original sources

PubMed2005

[Analysis of prognostic factors of and to establish a predictive model for patients with chronic severe hepatitis].

OBJECTIVE: To analyze prognostic factors of and to develop a prognostic model for patients with chronic severe hepatitis (CSH). METHODS: From December 1998 to October 2003, 385 in-patients being treated for chronic severe hepatitis were evaluated. The main clinical and laboratory variables were analyzed as predictive factors of survival with Cox univariate and multivariate regression models. RESULTS: The median survival time of this study group was 47 days. The survival rates at 1, 3, 6 months, 1-year, and 3 years were 66.2%, 32.9%, 26.9%, 22.9% and 17.7% respectively. Four prognostic factors were extracted using Cox's proportional hazard model; the prognostic index (PI) was calculated using the following formula consisting of these factors. PI = 0.016loge + 1.148 hepatic encephalopathy + 0.294loge (bilirubin micromol/L) - 0.826loge (prothrombin time activity). The model accurately predicted 3 months survival in an independent series of 84 patients with chronic severe hepatitis. CONCLUSION: The developed model is valuable in prognostic evaluation of chronic severe hepatitis and it may be useful to guide clinicians in selecting treatment methods for CSH.

Adolescent↗

PubMed2005

The in vitro skin irritation of chemicals: optimisation of the EPISKIN prediction model within the framework of the ECVAM validation process.

In view of the increasing need to identify non-animal tests able to predict acute skin irritation of chemicals, the European Centre for the Validation of Alternative Methods (ECVAM) focused on the evaluation of appropriate in vitro models. In vitro tests should be capable of discriminating between irritant (I) chemicals (EU risk: R38) and non-irritant (NI) chemicals (EU risk: "no classification"). Since major in vivo skin irritation assays rely on visual scoring, it is still a challenge to correlate in vivo clinical signs with in vitro biochemical measurements. Being particularly suited to test raw materials or chemicals with a wide variety of physical properties, in vitro skin models resembling in vivo human skin were involved in prevalidation processes. Among many other factors, cytotoxicity is known to trigger irritation processes, and can therefore be a first common event for irritants. A refined protocol (protocol 15min-18hours) for the EPISKIN model had been proposed for inclusion in the ECVAM formal validation study. A further improvement on this protocol, mainly based on a post-treatment incubation period of 42 hours (protocol 15min-42hours), the optimised protocol, was applied to a set of 48 chemicals. The sensitivity, specificity and accuracy with the MTT assay-based prediction model (PM) were 85%, 78.6% and 81.3% respectively, with a low rate of false negatives (12%). The improved performance of this optimised protocol was confirmed by a higher robustness (homogeneity of individual responses) and a better discrimination between the I and NI classes. To improve the MTT viability-based PM, the release of a membrane damage marker, adenylate kinase (AK), and of cytokines IL-1alpha and IL-8 were also investigated. Combining these endpoints, a simple two-tiered strategy (TTS) was developed, with the MTT assay as the first, sort-out, stage. This resulted in a clear increase in sensitivity to 95%, and a fall in the false-positive rate (to 4.3%), thus demonstrating its usefulness as a "decision-making" tool. The optimised protocol proved, both by its higher performances and by its robustness, to be a good candidate for the validation process, as well as a potential alternative method for assessing acute skin irritation.

Animal Testing Alternatives↗

PubMed2002

A novel area of predictive modelling: describing the functionality of beneficial microorganisms in foods.

Predictive microbiology generally focuses on the potential outgrowth of spoilage bacteria and foodborne pathogens in foods. Little attention has been paid to the biokinetics of beneficial foodgrade microorganisms, such as lactic acid bacteria. The latter is commonly used in the food fermentation industry, mainly for the in situ production of the antimicrobial lactic acid to extend the shelf life of the food. Furthermore, many strains show additional industrial potential as novel starter cultures since they produce functional metabolites, such as bacteriocins and exopolysaccharides. The production of these functional metabolites has been demonstrated during in vitro experiments, but in many cases these novel starter cultures seem to be less efficient when applied in a food system. A modelling approach may contribute to a better understanding of the tight relation between the food environment and bacterial functionality. Primary modelling can be applied to fit the experimental data concerning cell growth, sugar metabolism, and the production of functional metabolites for a given set of environmental conditions. This led to conclusions concerning the growth-associated production of bacteriocin and exopolysaccharides, the inactivation of these molecules when cell growth levels off, and a minimum cell concentration to trigger on bacteriocin production. Examples deal with the production of the bacteriocin sakacin K by the natural fermented sausage isolate Lactobacillus sakei CTC 494, and the production of heteropolysaccharides by the yoghurt starter culture Streptococcus thermophilus LY03. Secondary modelling of biokinetic parameters quantifies the production of bacteriocin and exopolysaccharides in function of environmental factors. As an example, the specific bacteriocin production by Lb. sakei CTC 494 decreases with increasing sodium chloride concentrations. Furthermore, since the assessment of functionality is frequently hampered by the nature of the food system, mathematical modelling techniques may help to predict the functional behaviour of novel lactic acid bacteria starter cultures in a food matrix, and hence quantify in situ production. For example, a model may simulate cell growth and exopolysaccharide production of S. thermophilus LY03 in a milk environment, where direct measurements are difficult to perform.

Bacteriocins↗

PubMed2002

A comprehensive and novel predictive modeling technique using detailed pathology factors in men with localized prostate carcinoma.

BACKGROUND: The purpose of the current study was to evaluate modeling strategies using sextant core prostate biopsy specimen data that would best predict biochemical control in patients with localized prostate carcinoma treated with permanent prostate brachytherapy (PPB). METHODS: One thousand four hundred seventy-seven patients underwent PPB between 1992 and 2000. The authors restricted analysis to those patients who had sextant biopsies (n = 1073). A central pathology review was undertaken on all specimens. Treatment consisted of PPB with either I-125 or Pd-103 prescribed to 144 Gy or 140 Gy, respectively. Two hundred twenty-eight patients (21%) received PPB in combination with external radiotherapy and 333 patients (31%) received neoadjuvant hormones. In addition to clinical stage, biopsy Gleason sum, and pretreatment prostate specific antigen (pretx-PSA), the following detailed biopsy variables were considered: mean percentage of cancer in an involved core; maximum percentage of cancer; mean primary and secondary Gleason grades; maximum Gleason grade (primary or secondary); percentage of cancer in the apex, mid, and base; percent of cores positive; maximum primary and secondary Gleason grades in apex, mid, and base; maximum percent cancer in apex, mid, and base; maximum Gleason grade in apex, mid, and base; maximum primary Gleason grade; and maximum secondary Gleason grade. In all, 23 biopsy variables were considered. Four modeling strategies were compared. As a base model, the authors considered the pretx-PSA, clinical stage, and biopsy Gleason sum as predictors. For the second model, the authors added percent of cores positive. The third modeling strategy was to use stepwise variable selection to select only those variables (from the total pool of 26) that were statistically significant. The fourth strategy was to apply principal components analysis, which has theoretical advantages over the other strategies. Principal components analysis creates component scores that account for maximum variance in the predictors. RESULTS: The median followup of the study cohort was 36 months (range, 6-92), and the Kattan modification of the American Society for Therapeutic Radiology and Oncology (ASTRO) definition was used to define PSA freedom from recurrence (PSA-FFR). The four models were compared in their ability to predict PSA-FFR as measured by the Somers D rank correlation coefficient. The Somers D rank correlation coefficients were then corrected for optimism with use of bootstrapping. The results for the four models were 0.32, 0.34, 0.37, and 0.39, respectively. CONCLUSIONS: The current study shows that the use of principal components analysis with additional pathology data is a more discriminating model in predicting outcome in prostate carcinoma than other conventional methods and can also be used to model outcome predictions for patients treated with radical prostatectomy and external beam.

Adult↗

PubMed2006

Validation of two prognostic models predicting outcome at two years after diagnosis in a new cohort of children with epilepsy: the Dutch Study of Epilepsy in Childhood.

PURPOSE: To validate two prognostic models for childhood-onset epilepsy designed to predict a terminal remission of <6 months at 2 years after diagnosis in children referred to the hospital. METHODS: A hospital-based cohort of children with newly diagnosed epilepsy was recruited and followed up for 2 years to validate previously developed models. One model was based on variables collected at intake, and the other was based on intake variables plus variables collected during the first 6 months of follow-up. The accuracy of both models was estimated by measuring the area under the receiver-operant-characteristic curves (ROC area). RESULTS: The ROC area of the model developed with intake variables was 0.69 [95% confidence interval (CI), 0.64-0.74] for the original cohort and 0.62 (95% CI, 0.55-0.69) for the validation cohort. The best combination of sensitivity and specificity for the original cohort was 61.6% and 69.1%, whereas it was 60.0% and 61.4% for the validation cohort. For the model with intake and 6-month variables combined, the ROC area was 0.78 (95% CI, 0.73-0.82) for the original cohort and 0.71 (95% CI, 0.64-0.78) for the validation cohort. The sensitivity and specificity were 72.6% and 73.1%, respectively, for the original cohort and 67.4% and 60.2%, respectively, for the validation cohort. CONCLUSIONS: Although both models predict outcome better than chance, they are insufficiently accurate to be of practical value. Both models performed marginally less well with the validation cohort than with the original cohort, but in both instances, the model based on intake and 6-month variables was more accurate.

Adolescent↗

PubMed1999

Preoperative prediction model of outcome after cholecystectomy for symptomatic gallstones.

BACKGROUND: After cholecystectomy for symptomatic gallstone disease 20%-30% of the patients continue to have abdominal pain. The aim of this study was to investigate whether preoperative variables could predict the symptomatic outcome after cholecystectomy. METHODS: One hundred and two patients were referred to elective cholecystectomy in a prospective study. Median age was 45 years; range, 20-81 years. A preoperative questionnaire on pain, symptoms, and history was completed, and the questions on pain and symptoms were repeated 1 year postoperatively. Preoperative cholescintigraphy and sonography evaluated gallbladder motility, gallstones, and gallbladder volume. Preoperative variables in patients with or without postcholecystectomy pain were compared statistically, and significant variables were combined in a logistic regression model to predict the postoperative outcome. RESULTS: Eighty patients completed all questionnaires. Twenty-one patients continued to have abdominal pain after the operation. Patients with pain 1 year after cholecystectomy were characterized by the preoperative presence of a high dyspepsia score, 'irritating' abdominal pain, and an introverted personality and by the absence of 'agonizing' pain and of symptoms coinciding with pain (P < 0.000001). In a constructed logistic regression model 15 of 18 predicted patients had postoperative pain (PVpos = 0.83). Of 62 patients predicted as having no pain postoperatively, 56 were pain-free (PVneg = 0.90). Overall accuracy was 89%. CONCLUSION: From this prospective study a model based on preoperative symptoms was developed to predict postcholecystectomy pain. Since intrastudy reclassification may give too optimistic results, the model should be validated in future studies.

Adult↗

PubMed1998

Predictive models of natural history in primary biliary cirrhosis.

Primary biliary cirrhosis is a slow, progressive disease. Although many years may elapse before asymptomatic primary biliary cirrhosis patients begin experiencing symptoms of liver disease, their overall survival is significantly lower than the normal population. The Mayo natural history model has been developed to depict patient survival in the absence of effective therapeutic intervention. Although there are a number of caveats in applying this model, it has been validated using external data sets and established as an accepted tool for clinical or research purposes. Furthermore, recent data suggest that the Mayo natural history model continues to provide useful, predictive information in the presence of ursodeoxycholic acid therapy, which has been shown to lower the serum bilirubin to the natural history model for patient survival. In addition to the natural history model for patient survival, mathematical models have been developed to describe histologic progression and development of esophageal varices.

Disease Progression↗

PubMed2003

A predictive model for the management of community-acquired pneumonia.

BACKGROUND: Understanding what determines the prognosis of community-acquired pneumonia (CAP) is especially important for decisions on hospitalization and antimicrobial therapy. The objective of the present study was to compare the predictability of mortality in our patients to that of the pneumonia patient outcomes research team (PORT) study. PATIENTS AND METHODS: Data of 320 patients admitted with CAP were retrospectively evaluated and classified according to the published scheme. RESULTS: One-month mortality was 14.4%; 1-year mortality was 27.8%, two-thirds from new episodes. Univariate logistic regression risk factors for the 1-month mortality rate included leukocytosis, anemia, hypoalbuminemia, elevated blood urea nitrogen, >or= two comorbidities, tachycardia, tachypnea, acidosis, stupor, age > 65 years and high serum lactic dehydrogenase. These variables, except the last two, plus pleural effusion and bilateral infiltration were also risk factors for 1-year mortality. In the multivariate models, eight of these factors were significant risk factors, four for 1-month mortality and six for 1-year mortality. Our model for prediction of 1-month mortality had a sensitivity of 65%, specificity of 95% and accuracy of 91%. CONCLUSION: Agreement between predictions by our model and the published model was considerable, showing that most patients in the low score groups should not have been hospitalized.

Adolescent↗

PubMed1998

Comparison of multiple prediction models for ambulation following spinal cord injury.

Few studies have properly compared predictive performance of different models using the same medical data set. We developed and compared 3 models (logistic regression, neural networks, and rough sets) in the in prediction of ambulation at hospital discharge following spinal cord injury. We used the multi-center Spinal Cord Injury Model System database. All models performed well and had areas under the receiver operating characteristic curve in the 0.88-0.91 range. All models had sensitivity, specificity, and accuracy greater than 80% at ideal thresholds. The performance of neural network and logistic regression methods was not statistically different (p = 0.48). The rough sets classifier performed statistically worse than either the neural network or logistic regression models (p-values 0.002 and 0.015 respectively).

Acute Disease↗

PubMed1990

Evaluation of a spectrum target prediction model in speech perception.

A model of a spectrum target prediction mechanism is proposed and evaluated by comparing predicted values with results of psychoacoustic experiments. When the trajectory of the cepstrally smoothed LPC spectrum is approximated by a second-order critically damped system, the proposed model can estimate target values using short-period spectrum sequences (50 ms) without being given the onset positions of the spectral transition. Additionally, this model decreases the length of transitional sounds and recovers vowel characteristics neutralized by coarticulation. Moreover, this model compensates for the transitions of syllables and extracts stable characteristics from syllable transitions. This model is applicable to coarticulation recovery in speech signal processing.

Humans↗

PubMed2000

Lower extremity function and subsequent disability: consistency across studies, predictive models, and value of gait speed alone compared with the short physical performance battery.

BACKGROUND: Although it has been demonstrated that physical performance measures predict incident disability in previously nondisabled older persons, the available data have not been fully developed to create usable methods for determining risk profiles in community-dwelling populations. Using several populations and different follow-up periods, this study replicates previous findings by using the Established Populations for the Epidemiologic Study of the Elderly (EPESE) performance battery and provides equations for the prediction of disability risk according to age, sex, and level of performance. METHODS: Tests of balance, time to walk 8 ft, and time to rise from a chair 5 times were administered to 4,588 initially nondisabled persons in the four sites of the EPESE and to 1,946 initially nondisabled persons in the Hispanic EPESE. Follow-up assessment for activity of daily living (ADL) and mobility-related disability occurred from 1 to 6 years later. RESULTS: In the EPESE, compared with those with the best performance (EPESE summary performance score of 10-12), the relative risks of mobility-related disability for those with scores of 4-6 ranged from 2.9 to 4.9 and the relative risk of disability for those with scores of 7-9 ranged from 1.5 to 2.1, with similar consistent results for ADL disability. The observed rates of incident disability according to performance level in the Hispanic EPESE agreed closely with rates predicted from models developed from the EPESE sites. Receiver operating characteristic curves showed that gait speed alone performed almost as well as the full battery in predicting incident disability. CONCLUSIONS: Performance tests of lower extremity function accurately predict disability across diverse populations. Equations derived from models using both the summary score and the gait speed alone allow for the estimation of risk of disability in community-dwelling populations and provide valuable information for estimating sample size for clinical trials of disability prevention.

Activities of Daily Living↗

PubMed2001

Establishment of in vitro cellular model predicting histocompatibility in allograft.

A novel in vitro cellular model producting recipient-donor histocompatibility in allograft was developed to select the donor validity. Fifteen couples of blood samples of donor and recipient in human BMT were examined using the model, and skin allograft in mice was performed to test the model. The results showed that the less the differences of histocompatibility evaluated by the model were, the later GVHR in human BMT occurred and the longer the survival time of skin allografts in mice. It was suggested that the model could be used to predict correctly histocompatibility between donor and recipient.

Animals↗

PubMed2026

Dissecting genetic variance structure and evaluating genomic prediction models for single-cross hybrids derived from Stiff Stalk and Non-Stiff Stalk maize heterotic groups.

The early 20th-century discovery of heterosis and the establishment of heterotic groups transformed maize (Zea mays L.) into a keystone of global agriculture. However, maize breeding faces two significant challenges: the gradual decline of general combining ability (GCA) variance within heterotic groups and the impracticality of testing all possible single crosses in the early stages of a breeding program. Here, we developed genomic best linear unbiased prediction (GBLUP)-based multikernel models, using additive and two alternative nonadditive genomic relationship matrices, to estimate the variance components associated with the general combining ability of Stiff Stalk (SS) and Non-Stiff Stalk (NSS) heterotic groups and the specific combining ability arising from their crosses. We further applied these models to predict the performance of untested single-cross combinations under varying levels of parental information. We showed that the SS and NSS groups retained significant GCA variance across traits in both early- and late-maturity groups. The SS group, in contrast, exhibited no detectable GCA variance in grain yield for the intermediate-flowering subset of hybrids, highlighting a limitation for future genetic improvement. Furthermore, our results showed that GBLUP-based multikernel models effectively identified superior hybrids when parental information was available. In the absence of this information, however, these models underperformed compared to covariance-based approaches. Both nonadditive matrices yielded similar results, indicating that they capture comparable genetic relationship patterns despite their distinct formulations. Overall, this study sheds light on the future use of US maize commercial germplasm and demonstrates how GBLUP-based multikernel models can improve the efficiency of hybrid breeding programs.

Zea mays↗

PubMed2006

Nonlinear finite element model predicts vertebral bone strength and fracture site.

STUDY DESIGN: A study on computed tomography (CT)-based finite element (FE) method that predicts vertebral strength and fracture site using human cadaveric specimens. OBJECTIVE: To evaluate the accuracy of the nonlinear FE method by comparing the predicted data with those of mechanical testing. SUMMARY OF BACKGROUND DATA: FE methods may predict vertebral strength and fracture site but the prediction has been difficult because of a complex geometry, elastoplasticity, and thin cortical shell of the vertebra. METHODS: FE models of the 12 thoracolumbar vertebral specimens were constructed. Nonlinear FE analyses were performed, and the yield load, the fracture load, the sites where elements failed, and the distribution of minimum principal strain were evaluated. A quasi-static uniaxial compression test for the same specimens was conducted to verify these analyses. RESULTS: The yield loads, fracture loads, minimum principal strains, and fracture sites of the FE prediction significantly correlated with those measured. CONCLUSIONS: Nonlinear FE model predicted vertebral strength and fracture site accurately.

Adult↗

PubMed2006

The Seattle Heart Failure Model: prediction of survival in heart failure.

BACKGROUND: Heart failure has an annual mortality rate ranging from 5% to 75%. The purpose of the study was to develop and validate a multivariate risk model to predict 1-, 2-, and 3-year survival in heart failure patients with the use of easily obtainable characteristics relating to clinical status, therapy (pharmacological as well as devices), and laboratory parameters. METHODS AND RESULTS: The Seattle Heart Failure Model was derived in a cohort of 1125 heart failure patients with the use of a multivariate Cox model. For medications and devices not available in the derivation database, hazard ratios were estimated from published literature. The model was prospectively validated in 5 additional cohorts totaling 9942 heart failure patients and 17,307 person-years of follow-up. The accuracy of the model was excellent, with predicted versus actual 1-year survival rates of 73.4% versus 74.3% in the derivation cohort and 90.5% versus 88.5%, 86.5% versus 86.5%, 83.8% versus 83.3%, 90.9% versus 91.0%, and 89.6% versus 86.7% in the 5 validation cohorts. For the lowest score, the 2-year survival was 92.8% compared with 88.7%, 77.8%, 58.1%, 29.5%, and 10.8% for scores of 0, 1, 2, 3, and 4, respectively. The overall receiver operating characteristic area under the curve was 0.729 (95% CI, 0.714 to 0.744). The model also allowed estimation of the benefit of adding medications or devices to an individual patient's therapeutic regimen. CONCLUSIONS: The Seattle Heart Failure Model provides an accurate estimate of 1-, 2-, and 3-year survival with the use of easily obtained clinical, pharmacological, device, and laboratory characteristics.

Adult↗

PubMed2006

Validation and re-evaluation of a discriminant model predicting anatomic suitability for biventricular repair in neonates with aortic stenosis.

OBJECTIVES: The purpose of this study was to validate and re-evaluate our previously reported scoring systems for predicting optimal management in neonates with aortic stenosis (AS). BACKGROUND: In 1991, we reported a multivariate discriminant equation and an ordinal scoring system for predicting which neonates with AS are suitable for biventricular repair and which are better served by single ventricle management. METHODS: Retrospective analysis was performed to: 1) validate our scoring systems in 89 additional neonates with AS and normal mitral valve area, 2) assess the effects of 5% measurement variation on predictive scores, 3) evaluate our cohort with the Congenital Heart Surgeons' Society scoring system, and 4) repeat the discriminant analysis on the basis of all 126 patients. RESULTS: The original scores each predicted outcome accurately in 68 patients (77%). Minor (5%) measurement variation changed the outcome predicted by the discriminant equation in 8 patients (9%) and by the threshold system in 13 patients (15%). The most accurate model for predicting survival with a biventricular circulation among the full cohort is: 10.98 (body surface area) + 0.56 (aortic annulus z-score) + 5.89 (left ventricular to heart long-axis ratio) - 0.79 (grade 2 or 3 endocardial fibroelastosis) - 6.78. With a cutoff of -0.65, outcome was predicted accurately in 90% of patients. CONCLUSIONS: Both of our original scoring systems are less accurate at predicting outcome than in our original analysis. Revised discriminant analysis yielded a model similar to our original equation that was 90% accurate at predicting survival with a biventricular circulation among neonates with AS and a mitral valve area z-score >-2.

Aortic Valve Stenosis↗

PubMedDate not supplied

Pelvic recurrence following resection of rectal cancer: a multivariate predictive model.

Local recurrence of rectal cancer (LR) after "curative" surgery is a major clinical problem, with a low resectability rate and a dismal prognosis. Prediction of LR might permit more targeted postoperative surveillance with earlier diagnosis of recurrent disease and might help in selecting the patients to be assigned to the most suitable adjuvant treatment protocol. To evaluate if a simple multivariate model could predict the LR and survival probability in the single case, we retrospectively evaluated 118 consecutive patients (63 males, 55 females; mean age 62 +/- 12 years) operated on for rectal cancer and followed up for a minimum of 4 years (range 51-111 months). Local recurrence rate was 28%, with a 6% of local + distant failure. Age and sex of patients, type of surgery, location of tumour in the rectum, size, morphology and grading of the tumour were all unrelated to the event under investigation. At Cox regression, the Dukes' stage and the postoperative radiotherapy were the only independent prognostic factors for LR (p < 0.001). The multivariate model was able to correctly reclassify the patients and predict local recurrence in 86.2% of the cases. Prevention of LR by adequate surgery and adjuvant therapy as well as its early detection offer the best prospect of improving the results of surgery for rectal cancer.

Adenocarcinoma↗

PubMed2004

Adsorption of organic vapors to air-dry soils: model predictions and experimental validation.

Soil/air equilibrium partitioning has an important impact on the environmental distribution and fate of many organic chemicals. Modeling approaches that cover this process commonly assume that sorption in soil only occurs in soil organic matter. However, many researchers have already shown thatthis is not even correctfor nonpolar compounds in air-dry soils. Here, we extend the existing data set on sorption in air-dry soils by using a large and very diverse set of organic compounds covering many different functional groups for two standard soils and for relative humidity between 50 and 90%. The experimental data presented here as well as those from the literature are then used to examine two different modeling approaches: one that only considers absorption in soil organic matter and one that also considers adsorption to mineral surfaces. The results clearly show that sorption in air-dry soil cannot be explained when adsorption to mineral surfaces is ignored. Only the model that considers both sorbing phases, organic matter and mineral surfaces, gives good agreement for all experimental data. Our model for predicting adsorption to mineral surfaces does not require any further fitting with experimental data; thus, it can readily be incorporated into existing fate models in order to improve the description of the soil/air partition process.

Adsorption↗