TNF-α (Tumor Necrosis Factor) is a proinflammatory cytokine that amplifies inflammatory response and promotes leukocyte recruitment. TNF-α is primarily produced by activated macrophages, among others, in response to infection, inflammation, or tissue damage. Given its central role in normal and abnormal immune responses, it is the target of several therapeutics, such as adalimumab and etanercept. TNF-α is also a prognostic and diagnostic biomarker associated with rheumatoid arthritis, Alzheimer's disease, multiple sclerosis, several kidney diseases, cancers, type 2 diabetes, sepsis, and others. Because TNF-α levels change dynamically during inflammatory responses, tools capable of sensitive and spatially resolved detection could enable improved monitoring of immune activity and disease progression. Single-walled carbon nanotubes (SWCNT) are cylindrical carbon lattices that emit distinct near-infrared bandgap photoluminescence. In this work, we evaluated three aptamer-based sensor constructs, plus an additional two iterations of one aptamer sequence, and two antibody-based sensor constructs for TNF-α that use SWCNT near-infrared photoluminescence signal transduction. Several, but not all, of these aptamer and antibody-based sensors sensitively and selectively detected TNF-α in human and bovine serum in a physiologically relevant range, and we found that their sensing was impacted by both passivation and incorporating an exogenous quencher onto the aptamer sequence. This study highlights the importance and challenges of translating previously-validated molecular recognition elements to new detection conditions, in this case on the surface of SWCNT and in challenging serum conditions. It also validated a lead sensor, the VR11-SWCNT aptamer construct with or without quencher chemistry and with surface passivation, that builds upon constructs that failed in serum. These results demonstrate a strategy toward synthesis of nanoscale optical sensors capable of detecting TNF-α. We anticipate that the sensors evaluated here will have utility in both the diagnosis and study of inflammation-driven chronic disease, while the sensor assessment framework will help drive the broader field of molecularly specific diagnostics.

Single-walled carbon nanotubes (SWCNT) are versatile building blocks for optical sensors. Their near infrared photoluminescent emission is determined by their chiral structure. Commercially available SWCNT are polydisperse, leading to spectral congestion and no ability single analyte sensing. Aqueous two-phase extraction (ATPE) is a technique to sort SWCNT structures for improved nanosensor performance and multiplexed sensing. To date, ATPE has not been demonstrated with bioconjugate-compatible surface chemistries. Here, we demonstrate improved and multiplexed cytokine sensing by ATPE sorting SWCNT species with amine-functionalized ssDNA. We show that this approach enables effective chirality sorting and direct antibody conjugations. We used sorted, functionalized SWCNT to simultaneously and specifically detect the inflammatory cytokines IL-12 and IL-6 with high sensitivity and selectivity, and improved robustness compared to unsorted SWCNT. We anticipate that such sensitive, multiplexed will be useful for implantable and bedside diagnostics of inflammatory processes while serving as a model for sensing other complex biological processes.

Polymeric mesoscale nanoparticles (MNPs) are 300 to 500 nm in diameter with a PEGylated surface that exhibit unique renal tropism as a function of that size range, as described in our prior work, with selectivity toward renal tubular epithelial cells. Despite their well-described therapeutic applications, we do not yet fully understand whether they maintain the unique mesoscale size range under various storage condition, nor do we understand the mechanisms of their internalization in renal tubular epithelial cells. Here, we investigated whether MNPs maintain their size when exposed to freeze-thaw cycles and storage outside their intended -20 °C condition. MNPs demonstrated negligible changes in size and polydispersity up to 4 freeze-thaw cycles, while we found an increase in size elevated temperature as a function of cargo loading. We then performed in vitro studies to evaluate MNP cellular uptake mechanisms using the human renal cell carcinoma tubular epithelial cell line 786-O treated with pharmacological inhibitors of uptake pathways. We found that MNP internalization is almost entirely prevented by dynamin inhibitors, while macropinocytosis inhibition also reduced uptake, suggesting that such standard nanoparticle uptake pathways are robust to the mesoscale size range.

Cardiovascular diseases are widespread causes of morbidity and mortality throughout the world. Nanomedicine is at the scale of biomolecular interfaces, the same scale as the development of plaques, thrombi, or misfolded proteins. A variety of nanomaterials, polymeric delivery tools, and engineered biologics have been developed to target and detect or treat the molecular causes of disease at the cellular level. In some cases, these nanostructures have been engineered to bypass immune recognition and deliver therapeutic cargoes. Targeted nanomaterial transport is also assistive technology for diagnostic imaging in cardiovascular disease. Selective localization of contrast agents for ultrasound, magnetic resonance, X-Ray, and optical imaging strategies enables disease-focused contrast. Continued nanomedicine research aims to produce highly specified and highly effective treatments for cardiovascular diseases, alongside enhancing point-of-care diagnostics. As these two sides of medicine progress, therapeutics and diagnostics meet at the nanoscale to create multifunctional image-enhancing and image-enhanced therapeutic tools against cardiovascular disease.

Semiconducting single-walled carbon nanotubes (SWCNT) are uniquely able to serve as sensor transducers for biological analytes in living organisms. This is in large part due to their tunable and photostable near-infrared fluorescence. In addition, SWCNT-based sensors can be multiplexed due to the library of unique structures that exhibit variant optical band gaps. In our work, we are using SWCNT transducers to detect the presence of cancer biomarkers and chemotherapeutic agents in live mice. In prior studies, we engineered an implantable semipermeable membrane-based SWCNT nanosensor device which could detect the ovarian cancer biomarker HE-4. We next engineered a hydrogel-based SWCNT nanosensor device which could be directly injected into live mice non-invasively via a dual-barrel syringe. We demonstrated that this hydrogel sensor device could detect the chemotherapeutic doxorubicin in live mice. Here, we evaluated the stability and optimized the near infrared fluorescence point-and-shoot spectroscopic detection of these sensors in vivo. In ongoing work, we are adapting this platform to detect breast cancer biomarkers, such as the estrogen receptor, and inflammatory cytokines in disease models. These studies are focused on engineering optoelectronic sensors to improve early detection of cancer. They are equally focused on engineering sensors to improve cancer treatment by monitoring the therapeutic window of anthracyclines.

Abstract Anthracycline chemotherapeutics are commonly used as frontline treatments for a wide array of cancers. However, their administration to patients results in substantial side effects, primarily cardiotoxicity, as well as myelosuppression and gastrointestinal toxicity. Current clinical management of such side effects is solely based on a lifetime dosage limit, which inhibits their anti-tumor efficacy. Many individualized factors, including age, family history of cardiovascular disease, treatment regimen, and other co-morbidities influence drug pharmacology. Despite this heterogeneity, there is no method for determining actual organ or tumor exposure to the treatment in an individual. Here, we developed an optical nanosensor array for four anthracyclines—doxorubicin, daunorubicin, epirubicin, and idarubicin. We used single-walled carbon nanotubes as the signal transducer due to their tunable near-infrared fluorescence. We screened twelve distinct ssDNA sequences paired with seven SWCNT (n,m ) species at increasing concentrations of each of the four anthracyclines. The spectral responses were then used to develop machine learning-based classification models to identify different anthracycline types and concentrations. The optimized extreme gradient boosting model was able to classify high levels of each anthracycline with 100% accuracy. Concentration-based classification by PCA was performed for each anthracycline, distinguishing low (≤ 5 µM) and high (> 5 µM) concentrations. Finally, we validated the sensor performance using synthetic urine and sweat. Our findings demonstrate the potential of carbon nanotube-based sensor array to measure the pharmacokinetics of anthracyclines in patients with the goal of enhancing anti-tumor efficacy and monitoring off-target toxicities.

Anthracycline chemotherapeutics are common chemotherapeutics that have substantial toxicities. There is substantial interpatient pharmacokinetic variability, though there is no method to quantify organ or tumor exposure. Here, we exposed an optical nanosensor array to detect each of four anthracyclines. We screened 12 ssDNA sequences paired with seven single-walled carbon nanotube (n,m) species against several concentrations of doxorubicin, daunorubicin, idarubicin, and epirubicin. Complex spectral responses were used to develop machine-learning-based classification models to quantify each anthracycline. The optimized extreme gradient boosting model classified high levels of each anthracycline with 100% accuracy. Principal component analysis distinguished low (≤5 μM) and high concentrations of each anthracycline. Finally, we validated selected ssDNA-(n,m) pair performance in synthetic urine and sweat. Our findings deliver a generalizable optical spectral fingerprinting methodology for hard-to-detect analytes. Their use in clinical biofluids portends the preclinical and potentially clinical pharmacokinetic measurement of anthracyclines to improve efficacy and reduce toxicities.

Monitoring dopamine in complex biological environments is essential for understanding neurological disorders, disease diagnosis, and it presents a unique chemical challenge. In this work, we rationally designed several single-walled carbon nanotube (SWCNT)-based near-infrared fluorescent sensors for dopamine using ssDNA aptamers as selective molecular recognition elements. The performance of three dopamine-selective aptamer-SWCNT hybrids and (GT)10-SWCNT constructs were evaluated and compared for their magnitude of response, sensitivity, and selectivity to dopamine. We performed these studies in buffer, in complex media with noradrenaline and serotonin, and in synthetic cerebrospinal fluid. We evaluated sensor constructs alone, with heat + divalent cation addition, and with four different molecular passivation agents. Ultimately, sensors passivated with bovine serum albumin (BSA) demonstrated strong selectivity for dopamine compared to chemically similar molecules and increased the magnitude of response compared to (GT)10-SWCNT. Concentration-response curves in PBS, in a serotonin and noradrenaline solution, and artificial cerebrospinal fluid (aCSF) revealed dynamic ranges between 30 and 200 nM, and we found that the response occurs within five minutes. Together, these results demonstrate that dopamine aptamer-SWCNT sensors enable more selective and robust optical detection in complex biological environments compared to ssDNA-SWCNT with no inherent biological selectivity for dopamine.

Single-walled carbon nanotubes (SWCNT) can serve as powerful transducers for optical nanobiosensors. As near-infrared (NIR) fluorophores, they are used for a wide variety of biological sensing and imaging applications in vitro and in vivo. Rational biosensor design relies on the use of biological recognition elements to detect the sensor's target. In rationally designing SWCNT nanobiosensors, the nanotubes are functionalized with a biological recognition motif, whose binding event induces a modulation in SWCNT fluorescence, which can be measured with NIR spectroscopy in a well plate or cuvette, in cells, or through tissue of live animals. In this review, the sensor design strategies and functional outcomes of rationally-designed optical SWCNT sensors are assessed that employ biological recognition elements for analyte specificity. The biomolecular recognition elements are divided into categories of proteins, peptides, or oligonucleotides, and assessed functionalization schemes, highlighting advances made in the fields of biomedical sensing and imaging through rational design. Finally, a perspective is offered on remaining challenges and future directions for the field of SWCNT optical sensor engineering and hurdles for translation to the clinic.

Diagnostic sensor development and clinical translation lag, in part, due to a lack of rapid, high-throughput screening methodologies. Optimization of high-throughput sensor development and deployment will likely help in expediting tools toward the clinic. To address this issue, we optimized high-throughput screening parameters using a near-infrared (NIR-II) plate reader attached to an external probe for in vivo testing. We assessed spectroscopy parameters to improve the speed and precision in screening a single-walled carbon nanotube (SWCNT)-based optical sensor. To do so, we assessed the appropriate well-plate specifications, including laser power, excitation wavelength, exposure time, and focal height parameters for SWCNT-based optical sensor development. We also used the plate reader to screen fluorescent SWCNTs that were endocytosed by a macrophage cell line. We then performed NIR probe spectroscopy to assess SWCNTs embedded within a methylcellulose hydrogel. Finally, we used the NIR probe to measure the SWCNT center wavelength and intensity from live immunocompetent mice. We anticipate that this framework may be broadly applicable to the development of near-infrared nanosensors with the potential for more rapid clinical diagnostic translation.

Monitoring dopamine in complex biological environments is essential for understanding neurological disorders and disease diagnosis, though it presents a unique chemical challenge. In this work, we rationally designed several single-walled carbon nanotube (SWCNT)-based near-infrared fluorescent sensors for dopamine using ssDNA aptamers as selective molecular recognition elements. The performance of three dopamine-selective aptamer-SWCNT hybrids and sensitive but non-selective (GT) 10 -SWCNT constructs were evaluated and compared for their magnitude of response, sensitivity, and selectivity to dopamine. We performed these studies in buffer, in complex media with noradrenaline and serotonin, and in synthetic cerebrospinal fluid. We evaluated sensor constructs alone, with heat + divalent cation addition, and with four different molecular passivation agents. Ultimately, sensors passivated with bovine serum albumin (BSA) demonstrated strong selectivity for dopamine relative to noradrenaline, serotonin, and ascorbic acid, with a greater magnitude of response compared to (GT) 10 -SWCNT. Concentration-response curves in PBS, in a serotonin and noradrenaline solution, and artificial cerebrospinal fluid (aCSF) revealed dynamic ranges between 30 and 200 nM, and we found that the response occurs within five minutes. Together, these results demonstrate that dopamine aptamer-SWCNT sensors enable more selective and robust optical detection in complex biological environments.

Near-infrared fluorescent semiconducting single-walled carbon nanotubes (SWCNT) are excellent transducers for bioanalytical sensor development. This is in part due to their tunable, photostable near-infrared photoluminescent signal which is not absorbed by biological tissues. In our prior studies, we have tuned SWCNT nanosensor response to selectively detect analytes of interest with both antibodies and aptamers. In this work, we developed a multiplexed antibody-based SWCNT nanosensor for pro-inflammatory cytokines using aqueous two-phase extraction (ATPE) separation of SWCNT ( n,m ) species and subsequent conjugation to IL-6 and IL-12 antibodies. Separately, we optimized bulk-species SWCNT complexed with anti-dopamine aptamers. Here, we immobilized both sets of sensors separately on glass culture dishes to study signaling using in vitro models of neural disease. The multiplexed cytokine sensor was used to detect toxin-induced inflammatory signaling in a co-culture model of neurodegenerative disease, while the dopamine sensor was used to detect potassium chloride-induced excitatory signaling from cultured neural cancer cells. In each, we demonstrated the ability to detect relevant signaling from a cell model of disease in vitro. Ongoing work is focused on optimizing these sensors for real-time signal monitoring and as implantable tools to study disease in live animal models of disease. These studies are focused on engineering a modular sensor platform which can be used to study neural disease models in vitro and in vivo with the long-term goals of developing better therapeutics and earlier diagnostics for neurodegenerative and related diseases.

Anthracycline chemotherapeutics are common chemotherapeutics that have substantial toxicities. There is substantial interpatient pharmacokinetic variability, though there is no method to quantify organ or tumor exposure. Here, we exposed an optical nanosensor array to detect each of four anthracyclines. We screened 12 ssDNA sequences paired with seven single-walled carbon nanotube (n,m) species against several concentrations of doxorubicin, daunorubicin, idarubicin, and epirubicin. Complex spectral responses were used to develop machine-learning-based classification models to quantify each anthracycline. The optimized extreme gradient boosting model classified high levels of each anthracycline with 100% accuracy. Principal component analysis distinguished low (<= 5 mu M) and high concentrations of each anthracycline. Finally, we validated selected ssDNA-(n,m) pair performance in synthetic urine and sweat. Our findings deliver a generalizable optical spectral fingerprinting methodology for hard-to-detect analytes. Their use in clinical biofluids portends the preclinical and potentially clinical pharmacokinetic measurement of anthracyclines to improve efficacy and reduce toxicities.

Abstract Inflammation is a hallmark of cancer development as well as a consequence of cancer progression. Chronic inflammatory pathways promote solid tumor growth, metastasis, and immune evasion. Indeed, chronic signaling of several inflammatory cytokines that lead to tumor formation include TNF-alpha, while persistent IL-6 signaling drives cell survival and proliferation as well as angiogenesis. IL-12, however, is a powerful anti-tumor cytokine, which promotes macrophage and T-cell cytotoxicity in addition to inhibiting angiogenesis. Despite their importance, it is difficult to model and monitor real-time inflammatory cytokine signaling in the tumor microenvironment during cancer initiation and progression. The ability to do so non-invasively in animal tumor models would allow for a better understanding of the key drivers of this important cancer hallmark, while the ability to do so in a patient could be a diagnostic and prognostic tool. To this end, we have engineered a multiplexed optical sensor platform for inflammatory cytokines TNF-alpha, IL-6, and IL-12. Individual sensors are synthesized from species-sorted single-walled carbon nanotubes (SWCNT), which exhibit tissue-transparent near-infrared fluorescence. For each cytokine, we non-covalently attached an antibody or ssDNA aptamer to SWCNT, allowing for molecularly-specific detection and a change in emission spectra upon biomarker binding. We found that sensor performance was quantitative with a limit of detection in the clinical range following inhibition of non-specific surface adsorption with passivation agent polymers and proteins. Further, we demonstrated the ability to detect IL-6 and IL-12, key pro-tumor and anti-tumor cytokines respectively, simultaneously in solution, as well as IL-6 excreted by macrophages in response to pro-inflammatory stimuli. Together, we anticipate further deployment of these inflammatory cytokine monitoring sensors in vivo and translation to patient diagnostic and prognostic profiling. Citation Format: Ryan M. Williams, Amelia Ryan, Syeda Rahman, Atara Israel. Engineered multiplexed optical sensors to detect inflammatory cytokines in the tumor microenvironment [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2026; Part 1 (Regular Abstracts); 2026 Apr 17-22; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2026;86(7 Suppl):Abstract nr 726.

Diagnostic sensor development and clinical translation lag, in part, due to a lack of rapid, high-throughput screening methodology. Optimization of high-throughput sensor development and deployment will likely help in expediting tools toward the clinic. To address this issue, we optimized high-throughput screening parameters using a near-infrared (NIR-II) plate reader attached to an external probe for in vivo testing. We assessed spectroscopy parameters to improve speed and precision in screening a single-walled carbon nanotube (SWCNT)-based optical sensor. To do so, we assessed the appropriate well plate specifications, including laser power, excitation wavelength, exposure time, and focal height parameters for SWCNT-based optical sensor development. We also used the plate reader to screen fluorescent SWCNT which were endocytosed by a macrophage cell line. We then performed NIR probe spectroscopy to assess SWCNT embedded within a methylcellulose hydrogel. Finally, we used the NIR probe to measure SWCNT center wavelength and intensity from live immunocompetent mice. We anticipate that this framework may be broadly applicable to the development of near infrared nanosensors with the potential for more rapid clinical diagnostic translation.

Tumor Necrosis Factor (TNF) is an inflammatory cytokine that regulates progression of cancer and other chronic diseases. TNF is produced by activated macrophages in response to infection, injury, apoptosis, angiogenesis, or other physiological processes. It is a relevant prognostic and diagnostic biomarker associated with inflammatory diseases such as Rheumatoid arthritis, Alzheimer's disease, Crohn’s disease, and other metabolic disorders. Detection and quantification of TNF in disease models can also be a powerful tool to understand the contributions of inflammatory processes to disease progression. Single-walled carbon nanotubes (SWCNT) are cylindrical graphene sheets that emit distinct bandgap fluorescence in the near infrared region which can be modulated by the local environment. In this work, we engineered seven molecularly-specific TNF sensors with transduction of binding using SWCNT and specific recognition using antibodies and nucleic acid aptamers. For each, we attempted to optimize binding conditions and signal transduction in buffer and serum, ultimately developing comparisons across each. We found that several sensor constructs demonstrated sensitive and specific detection in serum conditions, though not all did. We anticipate that one or multiple of these sensors will be combined with our previous injectable hydrogel nanosensor device to continuously detect TNF in animal models of disease, and that more broadly such a framework for evaluation of several sensor platforms for a target of interest may serve as a model for more translational sensor development.

Single-walled carbon nanotubes (SWCNT) are versatile building blocks for optical sensors. With proper surface modifications, SWCNT exhibit specific and selective responses to a wide variety of analytes. Their carbon lattice structure, denoted by chirality ( n,m ), determines the wavelength of their near infrared fluorescence emission. Commercially available SWCNT are polydisperse, containing many ( n,m ) species with overlapping fluorescence peaks, leading to spectral congestion. Nanotubes can be purified and sorted by their chiral structure through aqueous two-phase extraction (ATPE), a thermodynamically driven segregation. Chirality-pure SWCNT offer improved nanosensor performance and unique opportunities for multiplexed sensors. A persistent challenge in creating chirality-pure SWCNT sensors has been the difference in surface chemistries required for chirality isolation and those required for sensor functionalization. One possible approach may be to develop biosensors using SWCNT wrapped with ssDNA following chirality purification. However, these sensing approaches have been limited to analytes that can be detected with the same sequence used for chirality sorting. Further functionalization of SWCNT-DNA sorted through ATPE, such as biomolecular conjugation with analyte-specific antibodies, has yet to be realized. This gap is largely due to the potential of chemically modified DNA sequences, needed for functionalization, to disrupt chiral separation dynamic of ATPE. In this work, we demonstrated ATPE sorting of SWCNT using chemically-functionalized DNA (SWCNT-DNA-NH 2 ) for the first time. We evaluated the non-functionalized and aminated versions of DNA sequences known to be effective in ATPE sorting, finding similar chirality separation patterns for both – suggesting that ATPE is robust to chemically-modified ssDNA chirality recognition sequences. To demonstrate its utility, we performed antibody conjugation to the sorted SWCNT-DNA-NH 2 to produce highly specific chirality-purified sensors. Here, we used (6,5)- and (7,6)-purified nanosensors to demonstrate multiplexed sensing of cytokines IL-6 and IL-12 – two relevant biomarkers for inflammatory diseases. The sensors selectively discriminated between IL-6 and IL-12 through peak-specific wavelength shifts. Though inflammatory cytokines were chosen in this instance, the vast library of commercially available antibodies potentiates detection of many analytes that may be detected using this method.

Single walled carbon nanotubes (SWCNT) are novel nanomaterials with unique optical characteristics that can be leveraged to develop biosensors. Fluorescent semiconducting SWCNT are particularly interesting, since their emission wavelength and intensity are significantly sensitive to changes in surface chemistry, do not photobleach, and their near-infrared fluorescence penetrates biological media well. Thus, by functionalizing these nanotubes with biorecognition elements, analyte binding events can be translated to fluorescent changes. In this work, aptamers that specifically detect the neurotransmitter dopamine were used to wrap and solubilize SWCNTs. When the aptamer-SWCNT were exposed to dopamine, we observed a specific shift in characteristic fluorescent peaks, indicating the analyte is present in the sample. To enable a sensing platform that is capable of distinguishing between different neurotransmitters in a sample, passivation agents are being used to reduce non-specific interactions between the nanotube and the analyte. Different aptamer sequences were tested against dopamine and other neurotransmitters such as noradrenalin and serotonin. The specific detection of dopamine by this sensor could be then used to monitor levels of dopamine, which are altered in neurological and psychiatric disorders such as Parkinson and schizophrenia.

Cortisol is a hormone which regulates the body's response to stressors. Detection and monitoring of cortisol levels can provide information about physical and psychological health, thus it is essential to develop a sensor that can detect it in a sensitive manner. This study presents a biocompatible near-infrared fluorescent sensor, wherein single-walled carbon nanotubes (SWCNT) are functionalized with a cortisol-specific aptamer. We found this sensor was capable of detecting cortisol from 37.5 μg mL-1 to 300 μg mL-1 and that it was selective for cortisol compared to the similar molecule estrogen. Moreover, SWCNT functionalized with non-specific oligonucleotides did not exhibit a concentration-dependent response to cortisol, demonstrating the specificity provided by the aptamer sequence. The sensor also demonstrated the ability to detect cortisol in artificial cerebrospinal fluid. We anticipate that future optimization of this sensor will enable potential point-of-care or implantable device-based rapid detection of cortisol, with the potential for improving overall patient health and stress.

Anthracyclines are a class of chemotherapeutics used to treat many types of cancers. Their mechanisms of action include halting DNA synthesis through intercalation, or by inhibiting the enzymes topoisomerase I and II. This leads to mitochondrial dysfunction, production of reactive oxygen species (ROS), and causes cells to undergo oxidative stress leading to apoptosis. While this mechanism is beneficial for killing tumor cells, high rates of cardiotoxicity are associated with chemotherapy administration. There is a lifetime dose limit associated with anthracyclines in order to prevent such side effects. Particularly in cardiomyocytes, anthracyclines bind to cardiolipin, a phospholipid found in the respiratory chain, leading to subsequent cell death. While there are dosing limitations set in place for each anthracycline, there is no method to noninvasively monitor its accumulation in the body, and specifically in the heart. We therefore aimed to develop a single-walled carbon nanotube (SWCNT)-based sensor for the detection of four clinically used analogous anthracycline structures: daunorubicin, doxorubicin, epirubicin, and idarubicin. The benefits of using SWCNTs to develop an optical nanosensor include their stable fluorescence with no photobleaching or blinking in signal, and its ability to penetrate tissue easily being that they fluoresce in the near infrared region. SWCNTs were wrapped with 12 different ssDNA sequences and tested against each of the four anthracyclines at concentrations ranging from 0.1-100 µM for one hour at room temperature. Significant shifts in nanotube fluorescence center wavelength and intensity responses were evaluated, and principal component analysis (PCA) was used to determine if the responses can be differentiated based on ssDNA sequence, anthracycline/anthracycline concentration, and specific chirality response. While PCA was able to distinguish between and within some of these categories, it was not sufficient since it could not clearly separate all tested conditions from each other. We therefore decided to train machine learning models using the spectral responses of the sensors to correctly identify anthracycline presence and distinguish between them. We evaluated several supervised learning algorithms (support vector machine (SVM), decision tree (DT), random forest (RF) logistic regression, artificial neural network (ANN) and gradient boost), based on their F1-scores and ROC curves. Ongoing studies are focused on optimizing the models to improve their anthracycline predictions, to determine which sensor construct is best for detecting a specific anthracycline, and its limit of detection. Future use for these sensors includes encapsulating them into hydrogels to be used as a noninvasive sensor for in vivo anthracycline detection to better monitor its accumulation during the chemotherapy treatment process and mitigate any detrimental side effects.