The research centers around the concept of soft sensors—mathematical or statistical models that leverage process data and predictive algorithms to estimate product quality parameters dynamically. This approach contrasts sharply with hard sensors that directly measure physical properties but are limited by the complexity and invasiveness of pharmaceutical processes. By incorporating soft sensors, the pharmaceutical industry could achieve real-time, non-invasive monitoring of changes in the solid-state of oral dosage forms, mitigating risks associated with polymorphic transformations and ensuring quality-by-design principles throughout the manufacturing lifecycle.

Focusing on drug products exhibiting moderate to high risk levels due to solid-state variability, researchers employed diverse modeling frameworks tailored to different manufacturing unit operations. These encompassed population balance models, which effectively characterize particle size distributions; semi-empirical models designed to correlate process variables with solid-state transformations; and statistical correlation methods for uncovering intricate relationships between process parameters and quality outcomes. Key unit operations studied include wet granulation, fluidized bed drying, milling, and tablet compression, all critical stages where polymorphism and crystallinity can evolve profoundly, impacting drug performance.

A notable innovation highlighted in the study is the application of a population balance model within the wet granulation process. This model predicted particle size distribution by accounting for nucleation, growth, aggregation, and breakage phenomena as a function of granulation parameters. Complementing this, a smoothing splines model elegantly mapped polymorphic transitions relative to the liquid-to-solid ratio, providing nuanced insights into phase conversion dynamics. Such integrative modeling facilitates a comprehensive understanding of how subtle variations in wet granulator settings can cascade into meaningful changes in product solid-state attributes.

Moving to the fluidized bed dryer, the researchers utilized the Midilli-Kucuk empirical model to predict crucial drying metrics, including product temperature and moisture content. These variables are pivotal in controlling crystallinity, as residual moisture and thermal history directly influence solid-state transformations. The ability to forecast these parameters dynamically empowers process engineers to implement timely adjustments, ensuring consistency in the final product’s polymorphic form and crystalline content.

In the milling phase, addressing particle size reduction and crystalline integrity simultaneously presents a formidable challenge. Here, a population balance model tracked particle size evolution, capturing the attrition kinetics consistent with mechanical stress. Alongside, an exponential decay model described crystallinity loss over operational time, highlighting the deleterious impact of extended milling on solid-state properties. This dual-model approach underscores the delicate balance required to achieve desired particle metrics without compromising molecular structure.

The final manufacturing unit analyzed was the tablet press, where compression pressure exerts profound effects on tablet strength and polymorphism. By deploying statistical correlation analyses, the study identified clear relationships between applied pressure, tablet tensile strength, and polymorphic alterations. These insights are instrumental in optimizing compression parameters to maintain product integrity while meeting mechanical performance criteria.

Beyond the technical modeling, the study employs sensitivity analysis anchored in the partial rank correlation coefficient (PRCC) method. This rigorous statistical technique quantifies the influence of individual process variables on CQAs, illuminating critical control points. For instance, in the wet granulation context, the liquid-to-solid ratio emerged as a dominant factor driving polymorphic transformation, while impeller speed and wet massing time significantly modulated particle size distribution. Such findings provide a rational basis for prioritizing process parameters during manufacturing design and control.

An important dimension of this work is its regulatory context. The authors acknowledge that while direct FDA guidance specific to soft sensors is currently absent, these technologies align well with broader regulatory frameworks emphasizing advanced manufacturing practices, quality-by-design (QbD), and process analytical technology (PAT). Soft sensors exemplify model-driven control strategies that enhance data integrity and risk management, key pillars in ensuring pharmaceutical quality and patient safety. Their integration into production environments promises not only to satisfy regulatory expectations but also to propel innovation in process control paradigms.

Looking ahead, the researchers foresee expanding the scope of soft sensor applications through multiscale data integration, combining molecular-scale insights with process-level variables. This holistic approach could unlock unprecedented predictive capabilities, particularly for continuous manufacturing platforms where real-time adaptability is critical. Additionally, the work underscores the necessity for collaborative efforts spanning pharmaceutical companies, academic researchers, and regulatory agencies. Focus areas include model validation protocols, lifecycle management of predictive tools, and workforce training, all essential to realize the full potential of digitalized manufacturing landscapes.

The implications of this study extend far beyond academic inquiry, heralding a new era in which pharmaceutical manufacturing is governed by intelligent systems capable of preemptively identifying and mitigating risks related to solid-state transformations. As drug complexity increases and patient safety demands intensify, such advances will be indispensable. This research not only paves the way for more robust, reproducible, and scalable manufacturing processes but also fosters regulatory confidence in deploying cutting-edge technologies that ensure therapeutic efficacy and quality standards.

In summary, integrating soft sensors into pharmaceutical manufacturing provides a powerful avenue to predict, monitor, and control crucial quality attributes associated with crystallinity and polymorphism. By applying diverse modeling approaches tailored to specific unit operations and supplemented by comprehensive sensitivity analyses, this study offers a blueprint for future process control strategies that are both scientifically rigorous and practically implementable. The alignment with regulatory goals further reinforces the strategic value of these technologies in driving the evolution of pharmaceutical production into the digital age.

The comprehensive nature of this research signifies a major step toward operationalizing advanced sensor-based models for real-time quality assurance in drug production. As the pharmaceutical sector embraces digital transformation, the adoption of soft sensors could become a cornerstone technology, ensuring that solid oral dosage forms meet exacting standards from raw material processing to final tablet compression. The prospect of continuous, data-driven control and optimization holds the promise of safer, more effective medicines delivered consistently to patients worldwide.

Subject of Research: Not applicable
Article Title: Soft sensors to predict critical quality attributes and monitor crystallinity and polymorphism change in solid oral dosage manufacturing: case studies
News Publication Date: 5-Dec-2025
Web References: http://dx.doi.org/10.1007/s11705-025-2622-6
Image Credits: HIGHER EDUCATON PRESS

Keywords

Chemistry

The delicate respiratory physiology of preterm infants, especially those with extremely low birth weights, makes them particularly susceptible to rapid and harmful changes in arterial carbon dioxide levels. Traditional monitoring approaches often rely on intermittent arterial blood gas analyses, which provide only snapshot assessments of the infant’s respiratory status and expose neonates to repeated painful and risky procedures. By contrast, continuous transcutaneous monitoring offers a non-invasive, real-time window into the infant’s ventilatory state, potentially allowing for precise titration of ventilation parameters and immediate detection of derangements.

High-frequency ventilation (HFV), employed frequently in the neonatal intensive care unit (NICU) for VLBW infants, provides an effective mode of respiratory support by delivering rapid, small-volume breaths. While HFV can mitigate lung injury associated with conventional ventilation modes, it demands meticulous regulation of gas exchange to avoid fluctuating carbon dioxide levels. The dynamic environment of HFV accentuates the need for vigilant monitoring, as small adjustments may lead to substantial shifts in pCO₂, impacting cerebral blood flow and the risk of intraventricular hemorrhage.

The integration of continuous tCO₂ monitoring into NICU protocols emerges from the work of Bernatzky et al., whose recent study highlights its utility and safety profile. Their research elucidates how transcutaneous sensors, attached non-invasively to the infant’s skin, quantitatively measure carbon dioxide diffusion through the epidermis, producing reliable surrogate markers of arterial pCO₂. This approach provides continuous quantification without the interruptions inherent to blood sampling, enabling clinicians to respond proactively to trends rather than reactive snapshots.

Importantly, the study underscores that continuous tCO₂ values correlate strongly with arterial blood gas measurements, confirming the technology’s accuracy and clinical relevance. When implemented alongside HFV, this monitoring modality supports the fine-tuning of ventilatory support by providing immediate feedback on the infant’s respiratory carbon dioxide clearance. The continuous nature of the data stream allows for nuanced adjustments that preempt hypo- or hypercapnic episodes, fostering more stable physiological conditions critical for neurodevelopmental preservation.

Moreover, the reduction in blood sampling requirements is particularly salient in the fragile VLBW population, for whom cumulative blood loss can precipitate anemia and heighten the need for transfusions. By decreasing the dependency on repeated arterial punctures, continuous tCO₂ monitoring advances both patient comfort and safety. This less invasive method holds promise in improving not only clinical outcomes but also the overall neonatal intensive care experience for infants and families.

The technical advancements enabling reliable tCO₂ monitoring hinge on sensor calibration, skin site selection, and optimal device positioning to minimize artifact and ensure data fidelity. The system operates by heating the skin locally to increase capillary blood flow and CO₂ diffusion, with the sensor detecting partial pressure through electrochemical analyzers. Within the NICU setting, meticulous attention to sensor application and maintenance is paramount to prevent skin injury while ensuring consistent measurement accuracy.

Bernatzky and colleagues’ trial also delves into thresholds and alarm systems tailored to neonatal physiology, essential for integrating tCO₂ data into clinical workflow. Understanding the critical pCO₂ ranges for VLBW infants on HFV enables neonatologists to customize ventilation strategies, averting the extremes of hypocapnia, which can compromise cerebral perfusion, and hypercapnia, implicated in pulmonary vasoconstriction and acidosis. Real-time alerts can facilitate prompt interventions, reducing the incidence of potentially devastating complications.

The implications of adopting continuous tCO₂ monitoring extend beyond individual patient care to encompass broader healthcare systems. Decreasing the number of blood gas analyses per infant can alleviate laboratory workload and reduce healthcare costs without compromising the quality of care. Additionally, the non-invasive approach aligns with evolving standards emphasizing patient-centered care and minimal intervention in the NICU, an environment already fraught with sensory and procedural stressors.

Further research is anticipated to refine the application of tCO₂ monitoring technology, including its integration with automated ventilation systems and development of predictive algorithms that leverage continuous data to anticipate respiratory crises. These innovations may usher in an era of closed-loop ventilation control, where machine learning algorithms adjust support parameters autonomously based on real-time physiological inputs, potentially improving neonatal survival and neurodevelopmental trajectories.

As the neonatal community embraces this technology, education and training will be critical components to maximize its benefits. Neonatal nurses and physicians must become adept at interpreting continuous tCO₂ trends, recognizing the nuances of sensor data, and integrating these findings with other clinical parameters. Multidisciplinary collaboration will ensure that advances in monitoring translate seamlessly into enhanced patient outcomes.

This advance also raises important considerations regarding sensor design and comfort, particularly given the delicate and often compromised skin integrity of preterm infants. Continued innovation is necessary to develop sensors that minimize interference with thermoregulation and skin barrier function while delivering precise, continuous data, ensuring the technology’s widespread applicability and acceptance.

The broader neonatal research community eagerly awaits further randomized controlled trials to confirm the long-term benefits of tCO₂ monitoring in reducing morbidity associated with abnormal carbon dioxide levels. Preliminary data are compelling, signifying a potential reduction in intraventricular hemorrhage and chronic lung disease incidence through improved carbon dioxide management, which could markedly alter the landscape of neonatal care.

In summary, continuous transcutaneous CO₂ monitoring represents a critical leap forward in respiratory management of VLBW infants receiving high-frequency ventilation. By bridging the gap between invasive blood sampling and real-time physiological monitoring, this technology offers a sophisticated, patient-friendly approach to controlling pCO₂ levels. As clinical adoption expands, it holds promise to enhance neonatal outcomes, mitigate risks associated with current monitoring modalities, and shape the future of ventilatory support in the NICU.

The study by Bernatzky et al. embodies a significant stride toward optimizing the delicate balance of respiratory support in the most vulnerable neonatal patients. Continuous monitoring not only empowers clinicians with instant insight into respiratory dynamics but also aligns perfectly with the goal of minimizing procedural burden in these fragile infants. This advancement solidifies the role of innovative, technology-driven solutions in improving critical care neonatology.

Ultimately, continuous tCO₂ monitoring in VLBW infants fosters a new era of precision neonatal medicine, paving the way for improved survival rates, reduced complications, and better neurodevelopmental outcomes. The evolution from intermittent to continuous monitoring epitomizes the integration of technology with compassionate care, transforming neonatal respiratory management from reactive to proactive and predictive.

Subject of Research: Continuous transcutaneous carbon dioxide monitoring in very low birth weight (VLBW) infants on high-frequency ventilation.

Article Title: Continuous transcutaneous CO₂ monitoring in VLBW infants on high-frequency ventilation.

Article References:
Bernatzky, A., Fontana Stiglich, Y., Brandani, M. et al. Continuous transcutaneous CO₂ monitoring in VLBW infants on high-frequency ventilation. Pediatr Res (2025). https://doi.org/10.1038/s41390-025-04642-5

Image Credits: AI Generated

DOI: 17 December 2025

This study, published in the prestigious journal Neurophotonics, seeks to revolutionize the way scientists engage with Alzheimer’s research by facilitating real-time observation of plaque signals in freely moving mice. The researchers adapted fiber photometry, a method traditionally employed to capture neural activity, to monitor the fluorescent properties of a plaque-binding dye known as Methoxy-X04. What sets this approach apart is its minimal invasiveness, allowing for longitudinal studies without the typical ethical compromises associated with sacrificing the subjects.

Initial experiments revealed profound insights as the researchers employed flat optical fibers in anesthetized Alzheimer’s model mice, specifically the 5xFAD strain, known for its accelerated plaque accumulation. The results were striking: the fluorescence signals exhibited a robust correlation with the plaque density obtained through subsequent examination of brain tissue slices. This correlation was so strong that the researchers could train a machine learning model to classify animals based solely on their depth profiles of fluorescence, offering a glimpse into the potential for automated diagnostics.

The advancement did not stop there. The researchers moved on to test tapered optical fibers, which provide depth-resolved data from different brain regions. This innovation proved critical, as the tapered fibers were not only successful in detecting plaque distribution in brain tissue slices but also maintained their efficacy when implanted chronically in living mice. After injecting Methoxy-X04, researchers noted depth-specific fluorescence increases exclusively in the Alzheimer’s models, a clear indicator of amyloid plaque activity. This stark contrast emphasizes the ability of the technique to differentiate between pathological and healthy signaling in real-time, a feat rarely achieved in previous studies.

What makes this development particularly exciting is the operational flexibility afforded by the method, enabling monitoring in awake, freely moving animals. This capacity allows researchers to observe the natural behavior of the subjects while simultaneously tracking changes in amyloid plaque levels. As the studies progressed, it became evident that the fluorescence signals increased in a manner consistent with the expected trajectory of Alzheimer’s disease, implying not only that the technique is effective but also that it reflects the biological reality of the disease.

When compared to established methods, such as two-photon microscopy or optoacoustic tomography, the fiber-optic approach stands out by offering the advantage of long-term monitoring of deep brain regions without the need for anesthesia. This is a significant leap forward since existing techniques often involve invasive procedures that can alter physiological states and impede natural behavior, thereby compromising the quality of data collected. Furthermore, the simplicity and non-invasive nature of this technique could encourage widespread adoption among researchers studying neurodegenerative diseases.

The implications for therapeutic development are profound. By enabling scientists to monitor how potential treatments impact amyloid plaque accumulation in real time, this technology could significantly accelerate the pace of Alzheimer’s research. Given the complexities surrounding the disease and the challenges of clinical validation, developing a tool that promises continuous observation will usher in a new era in the pursuit of effective therapies.

In conclusion, the research conducted by the team at the University of Strathclyde and Italian Institute of Technology marks a significant milestone in Alzheimer’s disease research. By employing a fiber-optic approach combined with the fluorescent properties of Methoxy-X04, the researchers have not only developed a method for non-invasive monitoring of plaque signals but have also paved the way for future innovations. The potential applications of this technology extend beyond mere observation; it might one day help unearth novel therapeutic strategies and provide deeper insights into the mechanisms of disease progression.

As the field of Alzheimer’s research evolves, this study exemplifies the essential intersection of engineering and biology, highlighting how technological advances can provide solutions to some of the most pressing challenges faced in medical research today. Future endeavors will undoubtedly build on these foundational developments, ultimately driving forward our understanding of Alzheimer’s disease and, we hope, leading to more effective interventions.

Subject of Research: Non-invasive monitoring of amyloid plaques in Alzheimer’s disease using fiber photometry
Article Title: Depth-resolved fiber photometry of amyloid plaque signals in freely behaving Alzheimer’s disease mice
News Publication Date: 23-Sep-2025
Web References: https://www.spiedigitallibrary.org/journals/neurophotonics/volume-12/issue-03/035014/Depth-resolved-fiber-photometry-of-amyloid-plaque-signals-in-freely/10.1117/1.NPh.12.3.035014.full
References: N. Byron et al., “Depth-resolved fiber photometry of amyloid plaque signals in freely behaving Alzheimer’s disease mice,” Neurophotonics 12(3), 035014 (2025)
Image Credits: S. Sakata (University of Strathclyde); top-left image created in BioRender.

Keywords

Alzheimer disease, Amyloidosis, Photometry, Fiber optics, Fluorescence microscopy, Brain, Brain activity maps, Neuroimaging.

CO₂ retention, or hypercapnia, represents a critical concern in neonates, particularly those struggling with respiratory distress or chronic lung conditions. Elevated CO₂ levels can precipitate respiratory failure and complicate the clinical course, necessitating timely and accurate detection. Traditionally, assessing CO₂ retention involves arterial blood gas analysis or non-invasive capnography; both methods present challenges in the neonatal context due to invasiveness, intermittent sampling, or limited sensitivity. The introduction of a continuous, predictive model based on Bayesian inference promises an unprecedented level of monitoring finesse.

At the core of this research lies VIehl, Segar, and Vesoulis’s retrospective investigation into existing NICU datasets, applying Bayesian statistical frameworks to identify patterns indicative of CO₂ retention. By harnessing routinely collected parameters—likely including respiratory rate, tidal volume, oxygen saturation, and possibly transcutaneous CO₂ measurements—the algorithm dynamically gauges the probability that a neonate is experiencing dangerous CO₂ accumulation. This probabilistic approach inherently manages uncertainty, offering clinicians a nuanced risk assessment rather than a binary alert.

The methodology capitalized on the wealth of retrospective data accumulated in neonatal intensive care units, a treasure trove rife with physiological variability and diverse clinical scenarios. Bayesian models excel in situations with probabilistic dependencies, enabling the integration of prior knowledge with observed data. This capacity to update predictions as new information arrives is especially valuable in the volatile clinical trajectories common among critically ill neonates. The research team thus imbued the IVCO2 index with adaptability, making it responsive to real-time changes.

One of the most compelling aspects of this work is the compatibility of the IVCO2 index with existing medical monitoring devices. By feeding standard device outputs into the algorithm, the need for specialized hardware or intrusive procedures is eliminated, facilitating seamless integration in busy NICUs worldwide. This approach underscores a growing trend in neonatal care: the utilization of advanced computational tools to extract deeper insights from data already being generated, optimizing patient monitoring without additional burden.

The validation phase, crucial for translating theoretical models into clinical assets, demonstrated robust predictive performance across a heterogeneous neonatal cohort. The algorithm showed high sensitivity and specificity in flagging episodes of CO₂ retention retrospectively confirmed by blood gas analyses, providing evidence of its potential reliability and clinical utility. Such validation builds confidence that the IVCO2 index could serve as an early warning system, alerting caregivers to deteriorating respiratory status before overt clinical signs emerge.

Beyond immediate clinical benefits, the IVCO2 index paves the way for personalized medicine in neonatology. By adjusting predictions based on individual patient data and evolving conditions, it supports tailored respiratory management strategies. The capacity to predict CO₂ retention risk in near real-time allows clinicians to titrate ventilatory support judiciously, potentially reducing the risks associated with both under- and over-ventilation. This balance is critical for minimizing ventilator-induced lung injury and optimizing developmental outcomes.

The research team’s choice to employ a Bayesian framework reflects a thoughtful intersection of clinical needs and advanced statistics. Unlike deterministic models, Bayesian algorithms inherently embrace uncertainty and variability, essential attributes in neonatal physiology where rapid changes and unique patient characteristics abound. This mathematical approach aligns neatly with the nature of bedside decision-making, supplementing clinical intuition with rigorous, data-driven probabilities.

Moreover, this innovation resonates with the broader digital transformation sweeping through healthcare, characterized by artificial intelligence and machine learning integration into diagnostics and monitoring. The IVCO2 index embodies these trends within neonatal care, illustrating how statistical innovations can translate vast datasets into clinically meaningful tools. Its reliance on retrospective data also exemplifies ethical and efficient research practices, extracting maximum value from existing records while avoiding unnecessary patient risk.

Crucially, the successful validation lends itself to potential future developments, including prospective deployment in NICUs for real-time decision support. Integration with electronic health records and existing monitoring systems could usher in a new era where respiratory compromise is detected preemptively, triggering timely interventions that mitigate sequelae. Further research might also extend this approach to other critical physiological disturbances, amplifying its impact beyond CO₂ monitoring.

The visual representation of the IVCO2 index’s performance highlights distinct probability thresholds correlated with clinical CO₂ retention events. This stratification helps elucidate how varying risk levels could inform graduated clinical responses, from heightened surveillance to active therapeutic measures. Such granularity in risk assessment is paramount in neonatal settings where overreaction bears potential harm as much as neglect.

Importantly, the algorithm’s retrospective validation signifies a vital step, but prospective trials remain necessary to confirm efficacy in live clinical environments. The nuanced interplay between algorithmic predictions and clinician judgment will require exploration, ensuring that computational aids complement rather than complicate care. Nonetheless, this work lays a strong foundation for these next phases, presenting a compelling proof-of-concept.

Additionally, the IVCO2 index might offer insights into the pathophysiology of neonatal respiratory failure, revealing subtle trends not readily apparent through conventional monitoring. By illuminating early markers of CO₂ retention, it can enhance understanding of disease progression and response to therapy. This knowledge could catalyze novel therapeutic approaches, ultimately improving survival rates and quality of life in this vulnerable population.

In summary, the validation of the IVCO2 index signifies a remarkable advancement in neonatal respiratory monitoring, merging statistical innovation with clinical pragmatism. Its Bayesian predictive model capitalizes on existing data streams, reducing reliance on invasive testing while improving risk stratification of carbon dioxide retention. As neonatal care continues evolving toward precision medicine, tools like this will be indispensable in ensuring the safest possible start for the most fragile lives.

The study’s publication in the Journal of Perinatology heralds a new chapter in neonatal critical care technology, inviting further exploration and refinement. The coming years may see the IVCO2 index become a standard feature in NICUs, representing a triumph of interdisciplinary collaboration among neonatologists, bioengineers, and statisticians. Ultimately, this technology could save countless neonatal lives, underscoring the profound impact of innovative data science in medicine.

Subject of Research: Validation of a novel Bayesian predictive algorithm for detection of carbon dioxide retention in neonates using retrospective NICU data.

Article Title: Validation of a novel Bayesian predictive algorithm for detection of carbon dioxide retention using retrospective neonatal ICU data.

Article References:
Viehl, L.T., Segar, J.L. & Vesoulis, Z.A. Validation of a novel Bayesian predictive algorithm for detection of carbon dioxide retention using retrospective neonatal ICU data. J Perinatol (2025). https://doi.org/10.1038/s41372-025-02369-z

DOI: https://doi.org/10.1038/s41372-025-02369-z