Optimal Care for Kidney Health: Development of a Merit-based Incentive Payment System (MIPS) Value Pathway.

The Merit-based Incentive Payment System (MIPS) is a mandatory pay-for-performance program through the Centers for Medicare & Medicaid Services (CMS) that aims to incentivize high-quality care, promote continuous improvement, facilitate electronic exchange of information, and lower health care costs. Previous research has highlighted several limitations of the MIPS program in assessing nephrology care delivery, including administrative complexity, limited relevance to nephrology care, and inability to compare performance across nephrology practices, emphasizing the need for a more valid and meaningful quality assessment program. This article details the iterative consensus-building process used by the American Society of Nephrology Quality Committee from May 2020 to July 2022 to develop the Optimal Care for Kidney Health MIPS Value Pathway (MVP). Two rounds of ranked-choice voting among Quality Committee members were used to select among nine quality metrics, 43 improvement activities, and three cost measures considered for inclusion in the MVP. Measure selection was iteratively refined in collaboration with the CMS MVP Development Team, and new MIPS measures were submitted through CMS's Measures Under Consideration process. The Optimal Care for Kidney Health MVP was published in the 2023 Medicare Physician Fee Schedule Final Rule and includes measures related to angiotensin-converting enzyme inhibitor and angiotensin receptor blocker use, hypertension control, readmissions, acute kidney injury requiring dialysis, and advance care planning. The nephrology MVP aims to streamline measure selection in MIPS and serves as a case study of collaborative policymaking between a subspecialty professional organization and national regulatory agencies.

Keys to Driving Implementation of the New Kidney Care Models.

Contemporary nephrology practice is heavily weighted toward in-center hemodialysis, reflective of decisions on infrastructure and personnel in response to decades of policy. The Advancing American Kidney Health initiative seeks to transform care for patients and providers. Under the initiative's framework, the Center for Medicare and Medicaid Innovation has launched two new care models that align patient choice with provider incentives. The mandatory ESRD Treatment Choices model requires participation by all nephrology practices in designated Hospital Referral Regions, randomly selecting 30% of all Hospital Referral Regions across the United States for participation, with the remaining Hospital Referral Regions serving as controls. The voluntary Kidney Care Choices model offers alternative payment programs open to nephrology practices throughout the country. To help organize implementation of the models, we developed Driver Diagrams that serve as blueprints to identify structures, processes, and norms and generate intervention concepts. We focused on two goals that are directly applicable to nephrology practices and central to the incentive structure of the ESRD Treatment Choices and Kidney Care Choices: (1) increasing utilization of home dialysis, and (2) increasing the number of kidney transplants. Several recurring themes became apparent with implementation. Multiple stakeholders from assorted backgrounds are needed. Communication with primary care providers will facilitate timely referrals, education, and comanagement. Nephrology providers (nephrologists, nursing, dialysis organizations, others) must lead implementation. Patient engagement at nearly every step will help achieve the aims of the models. Advocacy with federal and state regulatory agencies will be crucial to expanding home dialysis and transplantation access. Although the models hold promise to improve choices and outcomes for many patients, we must be vigilant that they not do reinforce existing disparities in health care or widen known racial, socioeconomic, or geographic gaps. The Advancing American Kidney Health initiative has the potential to usher in a new era of value-based care for nephrology.

A Carbon Nanotube Sensor Array for the Label-Free Discrimination of Live and Dead Cells with Machine Learning.

Developing robust cell recognition strategies is important in biochemical research, but the lack of well-defined target molecules creates a bottleneck in some applications. In this paper, a carbon nanotube sensor array was constructed for the label-free discrimination of live and dead mammalian cells. Three types of carbon nanotube field-effect transistors were fabricated, and different features were extracted from the transfer characteristic curves for model training with linear discriminant analysis (LDA) and support-vector machines (SVM). Live and dead cells were accurately classified in more than 90% of samples in each sensor group using LDA as the algorithm. The recursive feature elimination with cross-validation (RFECV) method was applied to handle the overfitting and optimize the model, and cells could be successfully classified with as few as four features and a higher validation accuracy (up to 97.9%) after model optimization. The RFECV method also revealed the crucial features in the classification, indicating the participation of different sensing mechanisms in the classification. Finally, the optimized LDA model was applied for the prediction of unknown samples with an accuracy of 87.5-93.8%, indicating that live and dead cell samples could be well-recognized with the constructed model.

Nephrologist Performance in the Merit-Based Incentive Payment System.

RATIONALE & OBJECTIVE: The Merit-Based Incentive Payment System (MIPS) is the largest quality payment program administered by the Centers for Medicare & Medicaid Services. Little is known about predictors of nephrologist performance in MIPS. STUDY DESIGN: Cross-sectional analysis. SETTING & PARTICIPANTS: Nephrologists participating in MIPS in performance year 2018. PREDICTORS: Nephrologist characteristics: (1) participation type (individual, group, or MIPS alternative payment model [APM]), (2) practice size, (3) practice setting (rural, Health Professional Shortage Area [HPSA], or hospital based), and (4) geography (Census Division). OUTCOMES: MIPS Final, Quality, Promoting Interoperability, Improvement Activities, and Cost scores. Using published consensus ratings, we also examined the validity of MIPS Quality measures selected by nephrologists. ANALYTICAL APPROACH: Unadjusted and multivariable-adjusted linear regression models assessing the associations between nephrologist characteristics and MIPS Final scores. RESULTS: Among 6,117 nephrologists participating in MIPS in 2018, the median MIPS Final score was 100 (interquartile range, 94-100). In multivariable-adjusted analyses, MIPS APM participation was associated with a 12.5-point (95% CI, 10.6-14.4) higher score compared with individual participation. Nephrologists in large (355-4,294 members) and medium (15-354 members) practices scored higher than those in small practices (1-14 members). In analyses adjusted for practice size, practice setting, and geography, among individual and group participants, HPSA nephrologists scored 1.9 (95% CI, -3.6 to -0.1) points lower than non-HPSA nephrologists, and hospital-based nephrologists scored 6.0 (95% CI, -8.3 to -3.7) points lower than non-hospital-based nephrologists. The most frequently reported quality measures by individual and group participants had medium to high validity and were relevant to nephrology care, whereas MIPS APM measures had little relevance to nephrology. LIMITATIONS: Lack of adjustment for patient characteristics. CONCLUSIONS: MIPS APM participation, larger practice size, non-HPSA setting, and non-hospital-based setting were associated with higher MIPS scores among nephrologists. Our results inform strategies to improve MIPS program design and generate meaningful distinctions between practices that will drive improvements in care.

Size Discrimination of Carbohydrates via Conductive Carbon Nanotube@Metal Organic Framework Composites.

Traditional chemical sensing methodologies have typically relied on the specific chemistry of the analyte for detection. Modifications to the local environment surrounding the sensor represent an alternative pathway to impart selective differentiation. Here, we present the hybridization of a 2-D metal organic framework (Cu3(HHTP)2) with single-walled carbon nanotubes (SWCNTs) as a methodology for size discrimination of carbohydrates. Synthesis and the resulting conductive performance are modulated by both mass loading of SWCNTs and their relative oxidation. Liquid gated field-effect transistor (FET) devices demonstrate improved on/off characteristics and differentiation of carbohydrates based on molecular size. Glucose molecule detection is limited to the single micromolar concentration range. Molecular Dynamics (MD) calculations on model systems revealed decreases in ion diffusivity in the presence of different sugars as well as packing differences based on the size of a given carbohydrate molecule. The proposed sensing mechanism is a reduction in gate capacitance initiated by the filling of the pores with carbohydrate molecules. Restricting diffusion around a sensor in combination with FET measurements represents a new type of sensing mechanism for chemically similar analytes.

Breath Acetone Sensing Based on Single-Walled Carbon Nanotube-Titanium Dioxide Hybrids Enabled by a Custom-Built Dehumidifier.

Acetone is a metabolic byproduct found in the exhaled breath and can be measured to monitor the metabolic degree of ketosis. In this state, the body uses free fatty acids as its main source of fuel because there is limited access to glucose. Monitoring ketosis is important for type I diabetes patients to prevent ketoacidosis, a potentially fatal condition, and individuals adjusting to a low-carbohydrate diet. Here, we demonstrate that a chemiresistor fabricated from oxidized single-walled carbon nanotubes functionalized with titanium dioxide (SWCNT@TiO2) can be used to detect acetone in dried breath samples. Initially, due to the high cross sensitivity of the acetone sensor to water vapor, the acetone sensor was unable to detect acetone in humid gas samples. To resolve this cross-sensitivity issue, a dehumidifier was designed and fabricated to dehydrate the breath samples. Sensor response to the acetone in dried breath samples from three volunteers was shown to be linearly correlated with the two other ketone bodies, acetoacetic acid in urine and ß-hydroxybutyric acid in the blood. The breath sampling and analysis methodology had a calculated acetone detection limit of 1.6 ppm and capable of detecting up to at least 100 ppm of acetone, which is the dynamic range of breath acetone for someone with ketosis. Finally, the application of the sensor as a breath acetone detector was studied by incorporating the sensor into a handheld prototype breathalyzer.

Heterogeneous Growth of UiO-66-NH2 on Oxidized Single-Walled Carbon Nanotubes to Form "Beads-on-a-String" Composites.

In this work, we demonstrate a facile synthesis of UiO-66-NH2 metal-organic framework (MOF)/oxidized single-walled carbon nanotubes (ox-SWCNTs) composite at room temperature. Acetic acid (HAc) was used as a modulator to manipulate the morphology of the MOF in these composites. With a zirconium oxide cluster (Zr) to 2-aminoteraphthalate linker (ATA) 1:1.42 ratio and acetic acid modulator, we achieved predominately heterogeneous MOF growth on the sidewalls of CNTs. Understanding the growth mechanism of these composites was facilitated by conducting DFT calculations to investigate the interactions between ox-SWCNTs and the MOF precursors. The synthesized composites combine both microporosity of the MOF and electrical conductivity of the SWCNTs. Gas sensing tests demonstrated higher response for UiO-66-NH2/ox-SWCNT hybrid toward dry air saturated with dimethyl methylphosphonate (DMMP) vapor compared to oxidized single-walled carbon nanotubes (ox-SWCNTs) alone.

Machine learning-assisted calibration of Hg2+ sensors based on carbon nanotube field-effect transistors.

Nanomaterial-based electronic sensors have demonstrated ultra-low detection limits, down to parts-per-billion (ppb) or parts-per-trillion (ppt) concentrations. However, these extreme sensitivities also make them susceptible to signal saturation at higher concentrations and restrict their usage primarily to low concentrations. Here, we report machine learning techniques to create a calibration method for carbon nanotube-based field-effect transistor (FET) devices. We started with linear regression, followed by regression splines to capture the non-linearity in the data. Further improvements in model performance were obtained with regression trees. Finally we lowered the model variance and further boosted the model performance by introducing random forest. The resulting performance as measured by R2 was estimated to be 0.8260 using out-of-bag error. The methodology avoids saturation and extends the dynamic range of the nanosensors up to 12 orders of magnitude in analyte concentrations. Further investigations of the sensing mechanism include analysis of feature importance in each of the model we tested. Functionalized nanosensors demonstrate selective detection of Hg2+ ions with detection limits 10-14.36±0.78 M, and maintain calibration to concentrations as high as 1 mM. Application of machine learning techniques to investigate which features in the FET signal maximally correlate with concentration changes provide valuable insight into the carbon nanotube sensing mechanism and assist in the rational design of future nanosensors.

Telehealth for Home Dialysis in COVID-19 and Beyond: A Perspective From the American Society of Nephrology COVID-19 Home Dialysis Subcommittee.

The coronavirus disease 2019 (COVID-19) pandemic, technological advancements, regulatory waivers, and user acceptance have converged to boost telehealth activities. Due to the state of emergency, regulatory waivers in the United States have made it possible for providers to deliver and bill for services across state lines for new and established patients through Health Insurance Portability and Accountability Act (HIPAA)- and non-HIPAA-compliant platforms with home as the originating site and without geographic restrictions. Platforms have been developed or purchased to perform videoconferencing, and interdisciplinary dialysis teams have adapted to perform virtual visits. Telehealth experiences and challenges encountered by dialysis providers, clinicians, nurses, and patients have exposed health care disparities in areas such as access to care, bandwidth connectivity, availability of devices to perform telehealth, and socioeconomic and language barriers. Future directions in telehealth use, quality measures, and research in telehealth use need to be explored. Telehealth during the public health emergency has changed the practice of health care, with the post-COVID-19 world unlikely to resemble the prior era. The future impact of telehealth in patient care in the United States remains to be seen, especially in the context of the Advancing American Kidney Health Initiative.

Photoluminescence Response in Carbon Nanomaterials to Enzymatic Degradation.

Myeloperoxidase (MPO), a key enzyme released by neutrophils during inflammation, has been shown to catalyze the biodegradation of carbon nanomaterials. In this work, we perform photoluminescence studies on the MPO-catalyzed oxidation of graphene oxide (GO) and surfactant-coated pristine (6,5) single-walled carbon nanotubes (SWCNTs). The enzymatic degradation mechanism involves the introduction of defects, which promotes further degradation. Interestingly, the photoluminescence responses of GO and SWCNTs to enzymatic degradation are counterposed. Although the near-infrared (NIR) fluorescence intensity of SWCNTs at 998 nm is either unchanged or decreases depending on the surfactant identity, the blue fluorescence intensity of GO at 440 nm increases with the progression of oxidation by MPO/H2O2/Cl- due to the formation of graphene quantum dots (GQDs). Turn-on GO fluorescence is also observed with neutrophil-like HL-60 cells, indicative of potential applications of GO for imaging MPO activity in live cells. Based on these results, we further construct two ratiometric sensors using SWCNT/GO nanoscrolls by incorporating surfactant-wrapped pristine SWCNTs as the internal either turn-off (with sodium cholate (SC)) or reference (with carboxymethylcellulose (CMC)) sensor. The ratiometric approach enables the sensors to be more stable to external noise by providing response invariant to the absolute intensity emitted from the sensors. Our sensors show linear response to MPO oxidative machinery and hold the promise to be used as self-calibrating carbon nanomaterial-based MPO activity indicators.

Synthesis of Holey Graphene Nanoparticle Compounds.

Bulk-scale syntheses of sp2 nanocarbon have typically been generated by extensive chemical oxidation to yield graphite oxide from graphite, followed by a reductive step. Materials generated via harsh random processes lose desirable physical characteristics. Loss of sp2 conjugation inhibits long-range electronic transport and the potential for electronic band manipulation. Here, we present a nanopatterned holey graphene material electronically hybridized with metal-containing nanoparticles. Oxidative plasma etching of highly ordered pyrolytic graphite via previously developed covalent organic framework (COF)-5-templated patterning yields bulk-scale materials for electrocatalytic applications and fundamental investigations into band structure engineering of nanocomposites. We establish a broad ability (Ag, Au, Cu, and Ni) to grow metal-containing nanoparticles in patterned holes in a metal precursor-dependent manner without a reducing agent. Graphene nanoparticle compounds (GNCs) show metal-contingent changes in the valence band structure. Density functional theory investigations reveal preferences for uncharged metal states, metal contributions to the valence band, and embedding of nanoparticles over surface incorporation. Ni-GNCs show activity for oxygen evolution reaction in alkaline media (1 M KOH). Electrocatalytic activity exceeds 10,000 mA/mg of Ni, shows stability for 2 h of continuous operation, and is kinetically consistent via a Tafel slope with Ni(OH)2-based catalysis.

Measuring Quality in Kidney Care: An Evaluation of Existing Quality Metrics and Approach to Facilitating Improvements in Care Delivery.

BACKGROUND: Leveraging quality metrics can be a powerful approach to identify substantial performance gaps in kidney disease care that affect patient outcomes. However, metrics must be meaningful, evidence-based, attributable, and feasible to improve care delivery. As members of the American Society of Nephrology Quality Committee, we evaluated existing kidney quality metrics and provide a framework for quality measurement to guide clinicians and policy makers. METHODS: We compiled a comprehensive list of national kidney quality metrics from multiple established kidney and quality organizations. To assess the measures' validity, we conducted two rounds of structured metric evaluation, on the basis of the American College of Physicians criteria: importance, appropriate care, clinical evidence base, clarity of measure specifications, and feasibility and applicability. RESULTS: We included 60 quality metrics, including seven for CKD prevention, two for slowing CKD progression, two for CKD management, one for advanced CKD and kidney replacement planning, 28 for dialysis management, 18 for broad measures, and two patient-reported outcome measures. We determined that on the basis of defined criteria, 29 (49%) of the metrics have high validity, 23 (38%) have medium validity, and eight (13%) have low validity. CONCLUSIONS: We rated less than half of kidney disease quality metrics as highly valid; the others fell short because of unclear attribution, inadequate definitions and risk adjustment, or discordance with recent evidence. Nearly half of the metrics were related to dialysis management, compared with only one metric related to kidney replacement planning and two related to patient-reported outcomes. We advocate refining existing measures and developing new metrics that better reflect the spectrum of kidney care delivery.

Luminescence "Turn-On" Detection of Gossypol Using Ln3+-Based Metal-Organic Frameworks and Ln3+ Salts.

Gossypol (Gsp), a natural toxin concentrated in cottonseeds, poses great risks to the safe consumption of cottonseed products, which are used extensively throughout the food industry. In this work, we report the first luminescence "turn-on" sensors for Gsp using near-infrared emitting lanthanide (Ln3+) materials, including Ln3+ MOFs and Ln3+ salts. We first demonstrate that the Yb3+ photoluminescence of a Yb3+ MOF, Yb-NH2-TPDC, can be employed to selectively detect Gsp with a limit of detection of 25 µg/mL via a "turn-on" response from a completely nonemissive state in the absence of Gsp. The recyclability and stability of Yb-NH2-TPDC in the presence of Gsp was demonstrated by fluorescence spectroscopy and PXRD analysis, respectively. A variety of background substances present in practical samples that would require Gsp sensing, such as refined cottonseed oil, palmitic acid, linoleic acid, and α-tocopherol, did not interfere with the Yb3+ photoluminescence signal. We further identified that the "turn-on" of Yb-NH2-TPDC photoluminescence was due to the "antenna effect" of Gsp, as evidenced by spectroscopic studies and supported by computational analysis. This is the first report that Gsp can effectively sensitize Yb3+ photoluminescence. Leveraging this sensing mechanism, we demonstrate facile, highly sensitive, fast-response detection of Gsp using YbCl3·6H2O and NdCl3·6H2O solutions. Overall, we show for the first time that Ln3+-based materials are promising luminescent sensors for Gsp detection. We envision that the reported sensing approach will be applicable to the detection of a wide variety of aromatic molecules using Ln3+ compounds including MOFs, complexes, and salts.

Modification of Carbon Nitride/Reduced Graphene Oxide van der Waals Heterostructure with Copper Nanoparticles To Improve CO2 Sensitivity.

Carbon nitride/reduced graphene oxide (rGO) van der Waals heterostructures (vdWH) have previously shown exceptional oxygen sensitivity via a photoredox mechanism, making it a potential material candidate for various applications such as oxygen reduction reaction catalysis and oxygen sensing. In this work, the electronic structure of a carbon nitride/rGO composite is modified through the introduction of copper nanoparticles (NPs). When incorporated into a chemiresistor device, this vdWH displayed a newfound CO2 sensitivity. The effects of humidity and light were investigated and found to be crucial components for the CO2 sensitivity. Density functional theory calculations performed on a carbon nitride/copper NP@rGO model system found an enhanced stabilization of CO2 caused by H-bonds between the carbon nitride layer and chemisorbed CO2 on copper, pointing to the important role played by humidity. The synergetic effect between the carbon nitride layer interfaced with CuNP@rGO, in combination with humidity and light (395 nm) irradiation, is found to be responsible for the newfound sensitivity toward CO2.

Tetrahydrocannabinol Detection Using Semiconductor-Enriched Single-Walled Carbon Nanotube Chemiresistors.

Semiconductor-enriched single-walled carbon nanotubes (s-SWCNTs) have potential for application as a chemiresistor for the detection of breath compounds, including tetrahydrocannabinol (THC), the main psychoactive compound found in the marijuana plant. Herein we show that chemiresistor devices fabricated from s-SWCNT ink using dielectrophoresis can be incorporated into a hand-held breathalyzer with sensitivity toward THC generated from a bubbler containing analytical standard in ethanol and a heated sample evaporator that releases compounds from steel wool. The steel wool was used to capture THC from exhaled marijuana smoke. The generation of the THC from the bubbler and heated breath sample chamber was confirmed using ultraviolet-visible absorption spectroscopy and mass spectrometry, respectively. Enhanced selectivity toward THC over more volatile breath components such as CO2, water, ethanol, methanol, and acetone was achieved by delaying the sensor reading to allow for the desorption of these compounds from the chemiresistor surface. Additionally, machine learning algorithms were utilized to improve the selective detection of THC with better accuracy at increasing quantities of THC delivered to the chemiresistor.

Probing Ca2+-induced conformational change of calmodulin with gold nanoparticle-decorated single-walled carbon nanotube field-effect transistors.

Nanomaterials are ideal for electrochemical biosensors, with their nanoscale dimensions enabling the sensitive probing of biomolecular interactions. In this study, we compare field-effect transistors (FET) comprised of unsorted (un-) and semiconducting-enriched (sc-) single-walled carbon nanotubes (SWCNTs). un-SWCNTs have both metallic and semiconducting SWCNTs in the ensemble, while sc-SWCNTs have a >99.9% purity of semiconducting nanotubes. Both SWCNT FET devices were decorated with gold nanoparticles (AuNPs) and were then employed in investigating the Ca2+-induced conformational change of calmodulin (CaM) - a vital process in calcium signal transduction in the human body. Different biosensing behavior was observed from FET characteristics of the two types of SWCNTs, with sc-SWCNT FET devices displaying better sensing performance with a dynamic range from 10-15 M to 10-13 M Ca2+, and a lower limit of detection at 10-15 M Ca2+.

Holey Graphene Metal Nanoparticle Composites via Crystalline Polymer Templated Etching.

While graphene has sparked enormous research interest since its isolation in 2004, there has also been an interest in developing graphene composite materials that leverage graphene's extraordinary physical properties toward new technologies. Oxidative analogues such as graphene oxide and reduced graphene oxide retain many of the same properties of graphene. While these materials contain many functional moieties, defect formation through current oxidation methods is random which, despite reductive treatments, can never fully recover the properties of the starting material. In the interest of bridging the divide between these two sets of materials for composite materials, here we show a methodology utilizing 2-D covalent organic frameworks as templates for hole formation in graphene through plasma etching. The holes formed act as edge-only chemical handles while retaining a contiguous sp2 structure. Holey graphene structures generated act as autoreduction sites for small noble metal nanoparticles which return many of graphene's original electrical properties that can be used for functional composites. Composite materials here show 103 enhancement of the Raman signal of the underlying holey graphene as well as excellent calculated limits of detection in gas sensing of H2S (3 ppb) and H2 (10 ppm).

Ultrasound guidance for sural nerve conduction studies.

INTRODUCTION: Ultrasound can potentially identify nerves and guide recording and stimulating electrode placement for nerve conduction studies (NCS). This prospective study was performed to determine whether ultrasound guidance of sural NCS results in higher action potential amplitude, fewer stimuli required, lower stimulus strength required, and less pain experienced. METHODS: Fourteen healthy individuals underwent bilateral sural NCS, both with and without ultrasound guidance. Studies were separated by at least 48 h, and the order of testing was randomly assigned. RESULTS: Ultrasound guidance resulted in significantly fewer stimuli and lower stimuli strength required to obtain supramaximal responses (P < 0.01-0.03). Ultrasound guidance required significantly more time to perform than standard sural NCS (P < 0.01). There was no difference in sural nerve amplitude or pain rating between the 2 groups. DISCUSSION: Neuromuscular ultrasound can be used effectively to guide electrode placement during sural NCS. Muscle Nerve 59:705-707, 2019.

Growth of ZIF-8 on molecularly ordered 2-methylimidazole/single-walled carbon nanotubes to form highly porous, electrically conductive composites.

The combination of porosity and electrical conductivity in a single nanomaterial is important for a variety of applications. In this work, we demonstrate the growth of ZIF-8 on the surface of single-walled carbon nanotubes (SWCNTs). The growth mechanism was investigated and a molecularly ordered imidazole solvation layer was found to disperse SWCNTs and promote crystal growth on the sidewalls. The resultant ZIF-8/SWCNT composite demonstrates high microporosity and electrical conductivity. The ZIF-8/SWCNT composite displayed semiconducting electrical behavior and an increase in sensor sensitivity toward ethanol vapors versus pristine SWCNTs.

Machine-Learning Identification of the Sensing Descriptors Relevant in Molecular Interactions with Metal Nanoparticle-Decorated Nanotube Field-Effect Transistors.

Carbon nanotube-based field-effect transistors (NTFETs) are ideal sensor devices as they provide rich information regarding carbon nanotube interactions with target analytes and have potential for miniaturization in diverse applications in medical, safety, environmental, and energy sectors. Herein, we investigate chemical detection with cross-sensitive NTFETs sensor arrays comprised of metal nanoparticle-decorated single-walled carbon nanotubes (SWCNTs). By combining analysis of NTFET device characteristics with supervised machine-learning algorithms, we have successfully discriminated among five selected purine compounds, adenine, guanine, xanthine, uric acid, and caffeine. Interactions of purine compounds with metal nanoparticle-decorated SWCNTs were corroborated by density functional theory calculations. Furthermore, by testing a variety of prepared as well as commercial solutions with and without caffeine, our approach accurately discerns the presence of caffeine in 95% of the samples with 48 features using a linear discriminant analysis and in 93.4% of the samples with only 11 features when using a support vector machine analysis. We also performed recursive feature elimination and identified three NTFET parameters, transconductance, threshold voltage, and minimum conductance, as the most crucial features to analyte prediction accuracy.