Monitoring clinical trials based on predictive probability of significance

At a given point in a clinical trial, investigators may ask the question: "What is the likelihood of a significant result if the trial were continued?" One possible answer to this question is to examine a predictive probability of the significant difference with further patient accrual. [See, for example, Choi, Smith, and Becker (1985, Controlled Clinical Trials 6, 280-288).] This paper proposes and investigates the approach in trials for comparing the means of two normal populations. Two methods for calculating the predictive probability are examined. The results indicate that the predictive probability can be a useful conservative measure in monitoring trials.     

A comparison of methods for predicting vegetation type

Predictive modeling of vegetation patterns has wide application in vegetation science. In this paper I discuss three methods of predictive modeling using data from the alpine treeline ecotone as a case study. The study area is a portion of Glacier National Park, Montana. Parametric general linear models (GLM), artificial neural networks (ANN) and classification tree (CT) methods of predicting vegetation type are compared to determine the relative strength of each predictive approach and how they may be used in concert to increase understanding of important vegetation – environment relations. For each predictive method, vegetation type within the alpine treeline ecotone is predicted using a suite of environmental indicator variables including elevation, moisture potential, solar radiation potential, snow potential index, and disturbance history. Results from each of the predictive methods are compared against the real vegetation types to determine the relative accuracy of the methods.When the entire data field is examined (i.e., not evaluated by smaller spatial aggregates of data) the ANN procedure produces the most accurate predictions (=0.571); the CT predictions are the least accurate (=0.351). The predicted patterns of vegetation on the landscape are considerably different using the three methods. The GLM and CT methods produce large contiguous swaths of vegetation types throughout the study area, whereas the ANN method produces patterns with much more heterogeneity and smaller patches.When predictions are compared to reality at catchment scale, it becomes evident that the accuracy of each method varies depending upon the specific situation. The ANN procedure remains the most accurate method in the majority of the catchments, but both the GLM and PCT produce the most accurate classifications in at least one basin each.The variability in predictive ability of the three methods tested here indicates that there may not be a single best predictive method. Rather it may be important to use a suite of predictive models to help understand the environment – vegetation relationships. The ability to use multiple predictive methods to determine which spatial subunits of a landscape are outliers is important when identifying locations useful for climate change monitoring studies.     

Evaluation of Clinical and Immunological Markers for Predicting Virological Failure in a HIV/AIDS Treatment Cohort in Busia,Kenya

Background

In resource-limited settings where viral load (VL) monitoring is scarce or unavailable, clinicians must use immunological and clinical criteria to define HIV virological treatment failure. This study examined the performance of World Health Organization (WHO) clinical and immunological failure criteria in predicting virological failure in HIV patients receiving antiretroviral therapy (ART).

Methods

In a HIV/AIDS program in Busia District Hospital, Kenya, a retrospective, cross-sectional cohort analysis was performed in April 2008 for all adult patients (>18 years old) on ART for ≥12 months, treatment-naive at ART start, attending the clinic at least once in last 6 months, and who had given informed consent. Treatment failure was assessed per WHO clinical (disease stage 3 or 4) and immunological (CD4 cell count) criteria, and compared with virological failure (VL >5,000 copies/mL).

Results

Of 926 patients, 123 (13.3%) had clinically defined treatment failure, 53 (5.7%) immunologically defined failure, and 55 (6.0%) virological failure. Sensitivity, specificity, positive predictive value, and negative predictive value of both clinical and immunological criteria (combined) in predicting virological failure were 36.4%, 83.5%, 12.3%, and 95.4%, respectively.

Conclusions

In this analysis, clinical and immunological criteria were found to perform relatively poorly in predicting virological failure of ART. VL monitoring and new algorithms for assessing clinical or immunological treatment failure, as well as improved adherence strategies, are required in ART programs in resource-limited settings.     

Use of Terrestrial Invertebrates as Indicators of the Ecological Sustainability of Forest Management under the Montreal Process

Twelve nations involved in boreal or temperate forest management are committed to reporting on indicators under the Montreal Process as a mechanism for assessing progress towards sustainable forest management. For fauna, invertebrates are often considered too poorly known and diverse to include in sustainability indicator reporting. The alternative view, that no monitoring of sustainability can be considered adequate without inclusion of some invertebrate species, is espoused in this paper. The microhabitats of soil and litter, foliage and canopy, bark and branch, dead standing trees and coarse woody debris are highlighted as relevant in the context of determining the impacts of forest management on invertebrates and for selecting representative species. It is argued that a selection of those species from each of the key microhabitats that are restricted to later stages of succession should be monitored. This could be complemented by a selection of easily monitored species from a range of functional groups as a means of endeavouring to pick up adverse impacts not foreseen on the basis of present knowledge. In the longer term, habitat indices (developed from predictive models of fauna habitat) should be used to monitor the occurrence of indicator species across the broader landscape, rather than at specific sites where monitoring of species takes place. Most countries would be in a position to select indicator species and commence monitoring for some key microhabitats (e.g. soil and litter). However, further research is needed in many countries before indicator species can be selected for other key microhabitats (e.g. coarse woody debris).     

Capnography for assessing nocturnal hypoventilation and predicting compliance with subsequent noninvasive ventilation in patients with ALS

Background

Patients with amyotrophic lateral sclerosis (ALS) suffer from hypoventilation, which can easily worsen during sleep. This study evaluated the efficacy of capnography monitoring in patients with ALS for assessing nocturnal hypoventilation and predicting good compliance with subsequent noninvasive ventilation (NIV) treatment.

Methods

Nocturnal monitoring and brief wake screening by capnography/pulse oximetry, functional scores, and other respiratory signs were assessed in 26 patients with ALS. Twenty-one of these patients were treated with NIV and had their treatment compliance evaluated.

Results

Nocturnal capnography values were reliable and strongly correlated with the patients'' respiratory symptoms (R ² = 0.211–0.305, p = 0.004–0.021). The duration of nocturnal hypercapnea obtained by capnography exhibited a significant predictive power for good compliance with subsequent NIV treatment, with an area-under-the-curve value of 0.846 (p = 0.018). In contrast, no significant predictive values for nocturnal pulse oximetry or functional scores for nocturnal hypoventilation were found. Brief waking supine capnography was also useful as a screening tool before routine nocturnal capnography monitoring.

Conclusion

Capnography is an efficient tool for assessing nocturnal hypoventilation and predicting good compliance with subsequent NIV treatment of ALS patients, and may prove useful as an adjunctive tool for assessing the need for NIV treatment in these patients.     

A follow up study of cytomegalovirus infection in a group of Turkish renal transplant recipients using molecular assays

The clinical value of an in-house cytomegalovirus nested polymerase chain reaction (CMV-PCR) and a commercial molecular assay hybrid capture CMV DNA assay (HCA) was evaluated in monitoring a group of renal transplant patients for six months follow up. In this study, the sensitivity, specificity, positive predictive value, and negative predictive value of nested CMV DNA PCR assay and HCA at the beginning of the study were 70, 42.9, 46.7, 66.7, and 60, 78.6, 66.7, and 73.3% respectively. After six months, they were 80, 66.7, 80, 66.7 for CMV PCR and 73.3, 88.9, 91.7, 66.7% for HCA respectively. These results indicate that in monitoring and predicting CMV infections in renal transplant recipients, not only qualitative but also quantitative assays must be used together in order to decide the preemptive strategies.     

Natriuretic peptides for the detection of paroxysmal atrial fibrillation in patients with cerebral ischemia--the Find-AF study

Background and Purpose

Diagnosis of paroxysmal atrial fibrillation (AF) can be challenging, but it is highly relevant in patients presenting with sinus rhythm and acute cerebral ischemia. We aimed to evaluate prospectively whether natriuretic peptide levels and kinetics identify patients with paroxysmal AF.

Methods

Patients with acute cerebral ischemia were included into the prospective observational Find-AF study. N-terminal pro brain-type natriuretic peptide (NT-proBNP), brain-type natriuretic peptide (BNP) and N-terminal pro atrial-type natriuretic peptide (NT-proANP) plasma levels were measured on admission, after 6 and 24 hours. Patients free from AF at presentation received 7 day Holter monitoring. We prospectively hypothesized that patients presenting in sinus rhythm with NT-proBNP>median were more likely to have paroxysmal AF than patients with NT-proBNPResults281 patients were included, of whom 237 (84.3%) presented in sinus rhythm. 220 patients naïve to AF with an evaluable prolonged Holter ECG were analysed. In patients with NT-proBNP>median (239 pg/ml), 17.9% had paroxysmal AF in contrast to 7.4% with NT-proBNP<239 pg/ml (p = 0.025). The ratio of early (0 h) to late (24 h) plasma levels of NT-proBNP showed no difference between both groups. For the detection of paroxysmal atrial fibrillation, BNP, NT-proBNP and NT-proANP at admission had an area under the curve in ROC analysis of 0.747 (0.663–0.831), 0.638 (0.531–0.744) and 0.663 (0.566–0.761), respectively. In multivariate analysis, BNP was the only biomarker to be independently predictive for paroxysmal atrial fibrillation.

Conclusions

BNP is independently predictive of paroxysmal AF detected by prolonged ECG monitoring in patients with cerebral ischemia and may be used to effectively select patients for prolonged Holter monitoring.     

Sample size reestimation by Bayesian prediction

We review a Bayesian predictive approach for interim data monitoring and propose its application to interim sample size reestimation for clinical trials. Based on interim data, this approach predicts how the sample size of a clinical trial needs to be adjusted so as to claim a success at the conclusion of the trial with an expected probability. The method is compared with predictive power and conditional power approaches using clinical trial data. Advantages of this approach over the others are discussed.     

Prediction of Asthma Exacerbations in Children by Innovative Exhaled Inflammatory Markers: Results of a Longitudinal Study

Background

In asthma management guidelines the primary goal of treatment is asthma control. To date, asthma control, guided by symptoms and lung function, is not optimal in many children and adults. Direct monitoring of airway inflammation in exhaled breath may improve asthma control and reduce the number of exacerbations.

Aim

1) To study the use of fractional exhaled nitric oxide (FeNO) and inflammatory markers in exhaled breath condensate (EBC), in the prediction of asthma exacerbations in a pediatric population. 2) To study the predictive power of these exhaled inflammatory markers combined with clinical parameters.

Methods

96 asthmatic children were included in this one-year prospective observational study, with clinical visits every 2 months. Between visits, daily symptom scores and lung function were recorded using a home monitor. During clinical visits, asthma control and FeNO were assessed. Furthermore, lung function measurements were performed and EBC was collected. Statistical analysis was performed using a test dataset and validation dataset for 1) conditionally specified models, receiver operating characteristic-curves (ROC-curves); 2) k-nearest neighbors algorithm.

Results

Three conditionally specified predictive models were constructed. Model 1 included inflammatory markers in EBC alone, model 2 included FeNO plus clinical characteristics and the ACQ score, and model 3 included all the predictors used in model 1 and 2. The area under the ROC-curves was estimated as 47%, 54% and 59% for models 1, 2 and 3 respectively. The k-nearest neighbors predictive algorithm, using the information of all the variables in model 3, produced correct predictions for 52% of the exacerbations in the validation dataset.

Conclusion

The predictive power of FeNO and inflammatory markers in EBC for prediction of an asthma exacerbation was low, even when combined with clinical characteristics and symptoms. Qualitative improvement of the chemical analysis of EBC may lead to a better non-invasive prediction of asthma exacerbations.     

Evidence of Yersinia pestis DNA from fleas in an endemic plague area of Zambia

Background

Study protocols involving experimental animals often require the monitoring of different parameters not only in anesthetized, but also in free moving animals. Most animal research involves small rodents, in which continuously monitoring parameters such as temperature and heart rate is very stressful for the awake animals or simply not possible. Aim of the underlying study was to monitor heart rate, temperature and activity and to assess inflammation in the heart, lungs, liver and kidney in the early postoperative phase after experimental cardiopulmonary bypass involving 45 min of deep hypothermic circulatory arrest in rats. Besides continuous monitoring of heart rate, temperature and behavioural activity, the main focus was on avoiding uncontrolled death of an animal in the early postoperative phase in order to harvest relevant organs before autolysis would render them unsuitable for the assessment of inflammation.

Findings

We therefore set up a telemetry-based system (Data Science International, DSI?) that continuously monitored the rat's temperature, heart rate and activity in their cages. The data collection using telemetry was combined with an analysis software (Microsoft excel?), a webmail application (GMX) and a text message-service. Whenever an animal's heart rate dropped below the pre-defined threshold of 150 beats per minute (bpm), a notification in the form of a text message was automatically sent to the experimenter's mobile phone. With a positive predictive value of 93.1% and a negative predictive value of 90.5%, the designed surveillance and alarm system proved a reliable and inexpensive tool to avoid uncontrolled death in order to minimize suffering and harvest relevant organs before autolysis would set in.

Conclusions

This combination of a telemetry-based system and software tools provided us with a reliable notification system of imminent death. The system's high positive predictive value helped to avoid uncontrolled death and facilitated timely organ harvesting. Additionally we were able to markedly reduce the drop out rate of experimental animals, and therefore the total number of animals used in our study. This system can be easily adapted to different study designs and prove a helpful tool to relieve stress and more importantly help to reduce animal numbers.     

Entropies of negative incomes, Pareto-distributed loss, and financial crises

Health monitoring of world economy is an important issue, especially in a time of profound economic difficulty world-wide. The most important aspect of health monitoring is to accurately predict economic downturns. To gain insights into how economic crises develop, we present two metrics, positive and negative income entropy and distribution analysis, to analyze the collective "spatial" and temporal dynamics of companies in nine sectors of the world economy over a 19 year period from 1990-2008. These metrics provide accurate predictive skill with a very low false-positive rate in predicting downturns. The new metrics also provide evidence of phase transition-like behavior prior to the onset of recessions. Such a transition occurs when negative pretax incomes prior to or during economic recessions transition from a thin-tailed exponential distribution to the higher entropy Pareto distribution, and develop even heavier tails than those of the positive pretax incomes. These features propagate from the crisis initiating sector of the economy to other sectors.     

Impacts of Predictive Genomic Classifier Performance on Subpopulation-Specific Treatment Effects Assessment

We consider three (strong, moderate and mild) predictive biomarker scenarios with varying prevalence. As such, there is no treatment effect in the biomarker negative (g ^?) patient subpopulation. Relative to g ^?, there is a four-fold profound treatment effect in the biomarker positive (g ⁺) patient subpopulation, a strongly predictive scenario; a three-fold large g ⁺ subpopulation treatment effect, a moderately predictive scenario; and a two-fold modest g ⁺ subpopulation treatment effect, a mildly predictive scenario. In this paper, we focus on binary endpoint in prescribing treatment effect. Using a Breiman’s (Mach. Learn. 24:123–140, 1996) machine learning voting algorithm via a k-fold cross-validated approach applied by Freidlin et al. (Clin. Cancer Res. 16:691–698, 2010), a predictive biomarker may be developed. We consider development or discovery of a genomic biomarker using microarray gene expressions data in randomized controlled trials and validate the biomarker’s predictive performance in an independent data set.We investigate the classification performance characteristics of a binary genomic composite biomarker (expected to be predictive of treatment effects) including sensitivity, specificity, accuracy, positive predictive value and negative predictive value as a function of true sensitive prevalence. In doing so, we report the finding based on three representative tuning parameter sets with varying degree of rigor in their choices of the parameters ranging from highly rigorous, moderately rigorous to mildly rigorous. We articulate the rationales on the choices of tuning parameter sets. We also study the impacts of misclassification of genomic biomarker classifiers on their assessment of treatment effects in the positive and negative patient subpopulations, and all-comer patients.We elucidate via simulation studies on approaches to improve sensitivity when a biomarker is highly specific but poorly sensitive, a scenario that is most likely to lead to an incorrect test conclusion of an applicable significant treatment effect in a specific patient subpopulation or both positive and negative subpopulations. We explore when it will be beneficial to develop a binary predictive biomarker and conclude that hypothesis test inferences for the g ⁺ subpopulation treatment effect in the dual hypotheses setting (all-comer and g ⁺ alone) cannot be relied upon if the biomarker classifier is only highly specific and poorly sensitive or resulting in poor negative predictive value. The converse dual hypotheses (all-comer and g ^? alone) have the same concern, viz. highly sensitive and poorly specific or resulting in poor positive predictive value. In addition, we compare the predictive performance of a biomarker classifier between use of direct selection and selection from a candidate pool shedding favorable lights of direct selection approach where biological or mechanistic plausibility can be relied upon. Further research is needed if accurate classifier is required irrespective of prevalence level.     

Prospective inference of bioprocess cell viability through chemometric modeling of fluorescence multiway data

In this investigation, the fermentation step of a standard mammalian cell-based industrial bioprocess for the production of a therapeutic protein was studied, with particular emphasis on the evolution of cell viability. This parameter constitutes one of the critical variables for bioprocess monitoring since it can affect downstream operations and the quality of the final product. In addition, when the cells experiment an unpredictable drop in viability, the assessment of this variable through classic off-line methods may not provide information sufficiently in advance to take corrective actions. In this context, Process Analytical Technology (PAT) framework aims to develop novel strategies for more efficient monitoring of critical variables, in order to improve the bioprocess performance. Thus, in this work, a set of chemometric tools were integrated to establish a PAT strategy to monitor cell viability, based on fluorescence multiway data obtained from fermentation samples of a particular bioprocess, in two different scales of operation. The spectral information, together with data regarding process variables, was integrated through chemometric exploratory tools to characterize the bioprocess and stablish novel criteria for the monitoring of cell viability. These findings motivated the development of a multivariate classification model, aiming to obtain predictive tools for the monitoring of future lots of the same bioprocess. The model could be satisfactorily fitted, showing the non-error rate of prediction of 100%.     

Hybrid-EKF: Hybrid model coupled with extended Kalman filter for real-time monitoring and control of mammalian cell culture

In a decade when Industry 4.0 and quality by design are major technology drivers of biopharma, automated and adaptive process monitoring and control are inevitable requirements and model-based solutions are key enablers in fulfilling these goals. Despite strong advancement in process digitalization, in most cases, the generated datasets are not sufficient for relying on purely data-driven methods, whereas the underlying complex bioprocesses are still not completely understood. In this regard, hybrid models are emerging as a timely pragmatic solution to synergistically combine available process data and mechanistic understanding. In this study, we show a novel application of the hybrid-EKF framework, that is, hybrid models coupled with an extended Kalman filter for real-time monitoring, control, and automated decision-making in mammalian cell culture processing. We show that, in the considered application, the predictive monitoring accuracy of such a framework improves by at least 35% when developed with hybrid models with respect to industrial benchmark tools based on PLS models. In addition, we also highlight the advantages of this approach in industrial applications related to conditional process feeding and process monitoring. With regard to the latter, for an industrial use case, we demonstrate that the application of hybrid-EKF as a soft sensor for titer shows a 50% improvement in prediction accuracy compared with state-of-the-art soft sensor tools.     

Epstein-barr virus DNAemia monitoring for the management of post-transplant lymphoproliferative disorder

Background

Post-transplant lymphoproliferative disorder (PTLD) is a potentially fatal complication of allogeneic hematopoietic cell transplantation (HCT). Epstein-Barr virus (EBV) reactivation (detectable DNAemia) predisposes to the development of PTLD.

Model predictive control of glucose concentration in type I diabetic patients: An in silico trial

In this paper, the feedback control of glucose concentration in type I diabetic patients using subcutaneous insulin delivery and subcutaneous continuous glucose monitoring is considered. A recently developed in silico model of glucose metabolism is employed to generate virtual patients on which control algorithms can be validated against interindividual variability. An in silico trial consisting of 100 patients is used to assess the performances of a linear output feedback and a nonlinear state-feedback model predictive controller, designed on the basis of the in silico model. More than satisfactory results are obtained in the great majority of virtual patients. The experiments highlight the crucial role of the anticipative feedforward action driven by the meal announcement information. Preliminary results indicate that further improvements may be achieved by means of a nonlinear model predictive control scheme.     

β-D-glucan Surveillance with Preemptive Anidulafungin for Invasive Candidiasis in Intensive Care Unit Patients: A Randomized Pilot Study

Background

Invasive candidiasis (IC) is a devastating disease. While prompt antifungal therapy improves outcomes, empiric treatment based on the presence of fever has little clinical impact. Β-D-Glucan (BDG) is a fungal cell wall component detectable in the serum of patients with early invasive fungal infection (IFI). We evaluated the utility of BDG surveillance as a guide for preemptive antifungal therapy in at-risk intensive care unit (ICU) patients.

Methods

Patients admitted to the ICU for ≥3 days and expected to require at least 2 additional days of intensive care were enrolled. Subjects were randomized in 3∶1 fashion to receive twice weekly BDG surveillance with preemptive anidulafungin in response to a positive test or empiric antifungal treatment based on physician preference.

Results

Sixty-four subjects were enrolled, with 1 proven and 5 probable cases of IC identified over a 2.5 year period. BDG levels were higher in subjects with proven/probable IC as compared to those without an IFI (117 pg/ml vs. 28 pg/ml; p<0.001). Optimal assay performance required 2 sequential BDG determinations of ≥80 pg/ml to define a positive test (sensitivity 100%, specificity 75%, positive predictive value 30%, negative predictive value 100%). In all, 21 preemptive and 5 empiric subjects received systemic antifungal therapy. Receipt of preemptive antifungal treatment had a significant effect on BDG concentrations (p< 0.001). Preemptive anidulafungin was safe and generally well tolerated with excellent outcome.

Conclusions

BDG monitoring may be useful for identifying ICU patients at highest risk to develop an IFI as well as for monitoring treatment response. Preemptive strategies based on fungal biomarkers warrant further study.

Trial Registration

Clinical Trials.gov {"type":"clinical-trial","attrs":{"text":"NCT00672841","term_id":"NCT00672841"}}NCT00672841     

Monitoring Central Venous Catheter Resistance to Predict Imminent Occlusion: A Prospective Pilot Study

Background

Long-term central venous catheters are essential for the management of chronic medical conditions, including childhood cancer. Catheter occlusion is associated with an increased risk of subsequent complications, including bloodstream infection, venous thrombosis, and catheter fracture. Therefore, predicting and pre-emptively treating occlusions should prevent complications, but no method for predicting such occlusions has been developed.

Methods

We conducted a prospective trial to determine the feasibility, acceptability, and efficacy of catheter-resistance monitoring, a novel approach to predicting central venous catheter occlusion in pediatric patients. Participants who had tunneled catheters and were receiving treatment for cancer or undergoing hematopoietic stem cell transplantation underwent weekly catheter-resistance monitoring for up to 12 weeks. Resistance was assessed by measuring the inline pressure at multiple flow-rates via a syringe pump system fitted with a pressure-sensing transducer. When turbulent flow through the device was evident, resistance was not estimated, and the result was noted as “non-laminar.”

Results

Ten patients attended 113 catheter-resistance monitoring visits. Elevated catheter resistance (>8.8% increase) was strongly associated with the subsequent development of acute catheter occlusion within 10 days (odds ratio = 6.2; 95% confidence interval, 1.8–21.5; p <0.01; sensitivity, 75%; specificity, 67%). A combined prediction model comprising either change in resistance greater than 8.8% or a non-laminar result predicted subsequent occlusion (odds ratio = 6.8; 95% confidence interval, 2.0–22.8; p = 0.002; sensitivity, 80%; specificity, 63%). Participants rated catheter-resistance monitoring as highly acceptable.

Conclusions

In this pediatric hematology and oncology population, catheter-resistance monitoring is feasible, acceptable, and predicts imminent catheter occlusion. Larger studies are required to validate these findings, assess the predictive value for other clinical outcomes, and determine the impact of pre-emptive therapy.

Trial Registration

Clinicaltrials.gov NCT01737554     

Circulating tumor cells: highlight on practical implications

Circulating tumor cells (CTCs) are cells of presumed epithelial origin, whose prognostic and predictive value in metastatic cancer patients has recently been demonstrated. To date, the count of CTCs through the CellSearch? system represents a valid approach for monitoring disease status in patients with metastatic colorectal, breast, and prostate cancer; in these cancer types, a rise in the CTC count at any time during treatment predicts a poor outcome. Nevertheless, the clinical utility of monitoring CTC counts remains controversial, and what to do when CTC counts rise during therapy still remains an unanswered question. In this report, we suggest how to integrate CTC counts with their molecular characterization to better translate biologic information obtained on CTCs into daily clinical practice.     

Mammalian cell culture monitoring using in situ spectroscopy: Is your method really optimised?

In recent years, as a result of the process analytical technology initiative of the US Food and Drug Administration, many different works have been carried out on direct and in situ monitoring of critical parameters for mammalian cell cultures by Raman spectroscopy and multivariate regression techniques. However, despite interesting results, it cannot be said that the proposed monitoring strategies, which will reduce errors of the regression models and thus confidence limits of the predictions, are really optimized. Hence, the aim of this article is to optimize some critical steps of spectroscopic acquisition and data treatment in order to reach a higher level of accuracy and robustness of bioprocess monitoring. In this way, we propose first an original strategy to assess the most suited Raman acquisition time for the processes involved. In a second part, we demonstrate the importance of the interbatch variability on the accuracy of the predictive models with a particular focus on the optical probes adjustment. Finally, we propose a methodology for the optimization of the spectral variables selection in order to decrease prediction errors of multivariate regressions. © 2017 American Institute of Chemical Engineers Biotechnol. Prog., 33:308–316, 2017