CoChat
Sign Up
Cited by 214Spectroscopy Techniques in Biomedical and Chemical ResearchOvarian cancer diagnosis and treatmentAdvanced biosensing and bioanalysis techniques

Detection of ovarian cancer via the spectral fingerprinting of quantum-defect-modified carbon nanotubes in serum by machine learning

Open access
Kim, Mijin, Chen, Chen, Wang, Peng, Mulvey, Joseph J., Yang, Yoona, Wun, Christopher, Antman-Passig, Merav, Luo, Hong-Bin +7 more·2022·Nature Biomedical Engineering
DOI 10.1038/s41551-022-00860-yPMC PMC9108893
Open this paper in CoChat — free account

Abstract

Serum biomarkers are often insufficiently sensitive or specific to facilitate cancer screening or diagnostic testing. In ovarian cancer, the few established serum biomarkers are highly specific, yet insufficiently sensitive to detect early-stage disease and to impact the mortality rates of patients with this cancer. Here we show that a 'disease fingerprint' acquired via machine learning from the spectra of near-infrared fluorescence emissions of an array of carbon nanotubes functionalized with quantum defects detects high-grade serous ovarian carcinoma in serum samples from symptomatic individuals with 87% sensitivity at 98% specificity (compared with 84% sensitivity at 98% specificity for the current best clinical screening test, which uses measurements of cancer antigen 125 and transvaginal ultrasonography). We used 269 serum samples to train and validate several machine-learning classifiers for the discrimination of patients with ovarian cancer from those with other diseases and from healthy individuals. The predictive values of the best classifier could not be attained via known protein biomarkers, suggesting that the array of nanotube sensors responds to unidentified serum biomarkers.

AI snapshot

Written from the paper’s full text.

TL;DR

An array of quantum-defect-modified carbon nanotube nanosensors plus machine learning detects high-grade serous ovarian carcinoma from serum with 87% sensitivity at 98% specificity, exceeding the CA125/transvaginal-ultrasound benchmark of 84% sensitivity at 98% specificity. The sensor responses appear to capture known and unidentified serum biomarkers.

Key findings

  • The best SVM model, using five OCC-DNAs, achieved 93.9% sensitivity, 95.2% specificity and an F-score of 0.945 in 10-fold cross-validation for HGSOC versus all other disease plus healthy samples; an independent 54-sample test set gave 100% sensitivity at 98% specificity and an F-score of 0.978.
  • The optimized prediction model yielded 87% sensitivity at 98% specificity when PPV and sensitivity were equally weighted, compared with 84% sensitivity at 98% specificity for the current best clinical screening test using CA125 and transvaginal ultrasonography.
  • Conventional biomarkers underperformed: CA125 screening with a 50 U ml−1 cutoff gave 65.3% sensitivity, 88.3% specificity and an F-score of 0.621; logistic regression of additional biomarkers only marginally improved HGSOC prediction.
  • The sensor array responded quantitatively to CA125, HE4, YKL40 and bilirubin; SVR models gave r2 values of 0.719 for CA125, 0.55 for HE4 and 0.56 for YKL40, but biomarker-dependent features did not fully account for predictive value, suggesting unidentified biomarkers contribute.

Methods

  • OCC-DNA nanosensors were synthesized by introducing organic colour centres to (6,5)-SWCNTs via diazonium chemistry and encapsulating them with ssDNA; ten combinations were made from three OCCs and four DNA sequences, six with P < 0.10 were selected, and the best SVM used five.
  • The study used 269 serum samples from female patients: 215 for training/validation (49 HGSOC, 51 other gynaecologic diseases, 29 non-gynaecologic cancer, 25 remission including 7 HGSOC, 61 healthy donors) and 54 independent test samples (7 HGSOC, 5 other gynaecologic diseases, 32 non-gynaecologic diseases, 10 healthy). Spectra were collected at 2 h, 24 h and 72 h; features were dint, dint*, dwl and dwl*.
  • Logistic regression, decision tree, artificial neural networks, random forest and SVM were trained with Bayesian hyperparameter optimization and 10-fold cross-validation; SVM/SVR used a radial basis function kernel, and external validation used the independent sample set. PROBAST was used to assess risk of bias.

Limitations

  • Stated: specimens were obtained from symptomatic individuals at diagnosis, so prediction outcomes will be lower in the clinical screening setting with asymptomatic individuals; further studies are needed to evaluate pre-invasive and early-invasive disease.
  • Stated: the limited medical record of healthy donors and the enriched fraction of breast cancers in the non-HGSOC group of the test set may introduce systematic bias in participant selection and validation.
  • Stated and evident: the biomarkers and molecular interactions primarily responsible for the disease fingerprint are unknown and largely cannot be determined by current machine-learning methods; the independent test set was small (54 samples, 7 HGSOC), limiting external validation.

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.1038/s41551-022-00860-y

Cites

Statistics

Nothing to recompute.

Claims the paper makes

0 of 35 cited sources checked · 0 supported · 0 flagged

  • Findingown data

    The predictive values of the best classifier could not be attained via known protein biomarkers.

    “The predictive values of the best classifier could not be attained via known protein biomarkers, suggesting that the array of nanotube sensors responds to unidentified serum biomarkers.”
    Quote in the paperSubstantial claim
  • Findingown data

    The OCC-DNA nanosensor array with machine learning detects high-grade serous ovarian carcinoma in serum samples from symptomatic individuals with 87% sensitivity at 98% specificity.

    “Here we show that a ‘disease fingerprint’ acquired via machine learning from the spectra of near-infrared fluorescence emissions of an array of carbon nanotubes functionalized with quantum defects detects high-grade serous ovarian carcinoma in serum samples from symptomatic individuals with 87% sensitivity at 98% specificity”
    • Also stated in: § Results · plain
      “yielded 87% sensitivity at 98% specificity when PPV and sensitivity were equally weighted (β = 1).”
    Quote in the paperSubstantial claim
  • Findingown data

    SVM resulted in the best F-scores among the five machine learning algorithms tested.

    “We found that SVM resulted in the best F-scores among the 5 machine learning algorithms that we tested”
    Quote in the paperSubstantial claim
  • Findingnone given

    The sensor technology does not rely on antibody-based molecular recognition elements.

    “Fourth, because the technology does not rely on antibody-based molecular recognition elements, the sensors could be more robust than existing methods (Supplementary Fig. 4), enabling use in resource-limited settings and in technologies such as point-of-care and wearable/implantable devices42.”
    Quote in the paperSubstantial claim
  • Findingown data

    Each OCC-DNA nanosensor featured a pair of emission peaks that depend on the chemical nature of the OCC and DNA sequence.

    “Each OCC-DNA nanosensor featured a pair of emission peaks that depend on the chemical nature of the OCC and DNA sequence.”
    Quote in the paperSubstantial claim
  • May combine two claims. 23
  • Classified as probably not a substantial claim. 22

  • May combine two claims.

    Findingown data

    Principal component analysis of spectroscopic data showed healthy samples segregated into separate regions from disease samples, but HGSOC could not be separated from other disease conditions.

    “Similar to Fig. 2c, healthy samples showed the differentiable signatures from the disease samples, denoted by their segregation into separate regions in the PCA plot (Fig. 2d), but HGSOC could not be separated from the other disease conditions.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Findingown data

    The first two principal components accounted for 87.5% of the total variance.

    “The first two principal components accounted for 87.5% of the total variance”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    With the independent test set, the optimized SVM model resulted in 100% sensitivity at 98% specificity and an F-score of 0.978.

    “With this new sample set, the optimized SVM model resulted in 100% sensitivity at 98% specificity and an F-score of 0.978.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    At decreasing β from 3 to 0.2, sensitivity at 98% specificity systematically increased, although the improvement was statistically non-significant between μ values of 0.2, 0.5, 0.8 and 1.

    “At decreasing β from 3 to 0.2, sensitivity at 98% specificity systematically increased, although the improvement was statistically non-significant between μ values of 0.2, 0.5, 0.8 and 1 in this study”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

    Findingnone given

    The sensor technology can be used in a high-throughput fashion to facilitate the screening of large populations.

    “Third, this sensor technology can be used in a high-throughput fashion to facilitate the screening of large populations.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    Combining datasets obtained over multiple incubation durations could improve model performance, but the performance was only marginally better than using 2 h of serum incubation.

    “We found that combining datasets obtained over multiple incubation durations could improve the model performance, but the performance was only marginally better than using 2 h of serum incubation”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    Support vector machine models resulted in excellent sensitivity and specificity of HGSOC detection, with an accuracy approaching 95%, outperforming conventional serum biomarker-based identification.

    “Support vector machine models resulted in excellent sensitivity and specificity of HGSOC detection, with an accuracy approaching 95%, outperforming conventional serum biomarker-based identification.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    The best-performing prediction model was the sensor array combination of 4-N(C2H5)2*C T2C3T2C, 4-N(C2H5)2*(TAT)4, 3,4,5-F3*(TAT)4, 3,4,5-F3*(AT)15 and 3,4,5-F3*(GT)15, and yielded 87% sensitivity at 98% specificity when PPV and sensitivity were equally weighted (β = 1).

    “The best-performing prediction model was the sensor array combination of 4-N(C2H5)2*C T2C3T2C, 4-N(C2H5)2*(TAT)4, 3,4,5-F3*(TAT)4, 3,4,5-F3*(AT)15 and 3,4,5-F3*(GT)15, and yielded 87% sensitivity at 98% specificity when PPV and sensitivity were equally weighted (β = 1).”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    The averaged cross-validation score of the SVM model was 93.9% sensitivity, 95.2% specificity, and an F-score of 0.945 differentiating HGSOC from all other disease + healthy samples.

    “The averaged cross-validation score of the SVM model was 93.9% sensitivity, 95.2% specificity, and an F-score of 0.945 differentiating HGSOC from all other disease + healthy samples.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

    Findingreasoning

    The materials needed for the sensor cost approximately $5 per sample because of the small amount of OCC-DNAs needed for screening (<5 ng).

    “The materials needed for the sensor cost approximately $5 per sample because of the small amount of OCC-DNAs needed for screening (<5 ng).”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

    Findingnone given

    The OCC-DNAs remained colloidally stable when kept at 4 °C until used (up to 6 months).

    “The OCC-DNAs were kept at 4 °C until used (up to 6 months) as the OCC-DNAs remained colloidally stable.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    The best-performing HGSOC prediction model included spectroscopic variables that were not sensitive to the known biomarkers, and their relative importance was much more significant than the biomarker-related variables.

    “It is important to note that the best-performing HGSOC prediction model (Fig. 3d) included the spectroscopic variables that were not sensitive to the known biomarkers, and their relative importance was much more significant than the biomarker-related variables (Extended Data Fig. 7).”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Findingown data

    Feature importance of the prediction models shows which spectral parameters most impacted the model performance using an ablation study.

    “Feature importance of the prediction models shows which spectral parameters most impacted the model performance using an ablation study.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Premisecitation

    In ovarian cancer, established serum biomarkers are highly specific but insufficiently sensitive to detect early-stage disease and to impact mortality.

    “In ovarian cancer, the few established serum biomarkers are highly specific, yet insufficiently sensitive to detect early-stage disease and to impact the mortality rates of patients with this cancer.”
    • Also stated in: § Abstract · plain
      “As established serum biomarkers do not sufficiently represent the ovarian cancer disease state and are not sensitive enough to achieve precise early diagnosis to benefit survival rates10”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites [10] Menon, U. et al. Ovarian cancer population screening and mortality after long-term follow-up in the UK Collaborative Trial of Ovarian Cancer Screening (UKCTOCS): a randomised controlled trial. Lancet 397, 2182–2193 (2021). source

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    Perception-based approaches have been used to classify various disease conditions on the basis of patterns in DNA methylation, volatile organic compounds, small metabolites, and imaging data.

    “Perception-based approaches have been used to classify various disease conditions on the basis of different patterns in methylation of DNA sequencing13, volatile organic compounds using electronic noses14, small metabolites using mass spectrometry15, and image analysis of pathology, computerized tomography scans and magnetic resonance imaging data16,17.”
    Quote in the paperNot a substantial claim0 of 3 sources checked
    • cites [13] Liu, M. C. et al. Sensitive and specific multi-cancer detection and localization using methylation signatures in cell-free DNA. Ann. Oncol. 31, 745–759 (2020). source

      Not checked in the free run.

    • cites [14] Hao, Y. et al. Detection of volatile organic compounds in breath as markers of lung cancer using a novel electronic nose. Proc. IEEE Sens. 2, 1333–1337 (2003).

      Not checked in the free run.

    • cites [15] Zhang, J. et al. Nondestructive tissue analysis for ex vivo and in vivo cancer diagnosis using a handheld mass spectrometry system. Sci. Transl. Med. 9, eaan3968 (2017). source

      Not checked in the free run.

  • Classified as probably not a substantial claim. May combine two claims.

    Premisecitation

    Longitudinal CA125 measurements combined with transvaginal ultrasonography improve positive predictive value for ovarian cancer detection, but the benefit of screening is still elusive.

    “Longitudinal CA125 measurements combined with transvaginal ultrasonography, result in improved PPV for ovarian cancer detection, but the benefit of screening is still elusive9,10.”
    Quote in the paperSubstantial claim0 of 2 sources checked
    • cites [9] Jacobs, I. J. et al. Ovarian cancer screening and mortality in the UK Collaborative Trial of Ovarian Cancer Screening (UKCTOCS): a randomised controlled trial. Lancet 387, 945–956 (2016). source

      Not checked in the free run.

    • cites [10] Menon, U. et al. Ovarian cancer population screening and mortality after long-term follow-up in the UK Collaborative Trial of Ovarian Cancer Screening (UKCTOCS): a randomised controlled trial. Lancet 397, 2182–2193 (2021). source

      Not checked in the free run.

  • May combine two claims.

    Premisecitation

    Semiconducting single-walled carbon nanotubes exhibit intrinsic near-infrared fluorescence with environmental responsivity down to the single-molecule level.

    “Semiconducting single-walled carbon nanotubes (SWCNTs) exhibit intrinsic near-infrared fluorescence19 with environmental responsivity down to the single-molecule level20.”
    Quote in the paperSubstantial claim0 of 2 sources checked
    • cites [19] Bachilo, S. M. et al. Structure-assigned optical spectra of single-walled carbon nanotubes. Science 298, 2361 (2002). source

      Not checked in the free run.

    • cites [20] Cognet, L. et al. Stepwise quenching of exciton fluorescence in carbon nanotubes by single-molecule reactions. Science 316, 1465 (2007). source

      Not checked in the free run.

  • May combine two claims.

    Premisecitation

    Fluorescence modulation of single-walled carbon nanotubes is caused by several mechanisms, including Fermi level shifting through modulation of the immediate redox environment, exciton disruption in response to binding events, and solvatochromic shifting due to perturbation of the local dielectric environment.

    “The fluorescence modulation of SWCNTs is caused by several mechanisms, including Fermi level shifting through modulation of the immediate redox environment and exciton disruption in response to binding events, which change SWCNT intensity, and solvatochromic (wavelength) shifting due to perturbation of the local dielectric environment, including shifts due to modulation of the local electrostatic”
    Quote in the paperSubstantial claim0 of 2 sources checked
    • cites [24] Roxbury, D., Jena, P. V., Shamay, Y., Horoszko, C. P. & Heller, D. A. Cell membrane proteins modulate the carbon nanotube optical bandgap via surface charge accumulation. ACS Nano 10, 499–506 (2016). source

      Not checked in the free run.

    • cites [40] Heller, D. A. et al. Peptide secondary structure modulates single-walled carbon nanotube fluorescence as a chaperone sensor for nitroaromatics. Proc. Natl Acad. Sci. USA 108, 8544 (2011). source

      Not checked in the free run.

  • May combine two claims.

    Interpretationown data

    The sensor elements' responses to known biomarkers did not account for all of the predictive value, suggesting that unknown biomarkers play an important role in the differentiation of HGSOC by the sensors.

    “but they did not account for all of the predictive value, suggesting that unknown biomarkers play an important role in the differentiation of HGSOC by the sensors.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Interpretationown data

    The independent test set values are consistent with cross-validation scores and gave a similar ROC curve, indicating that the model did not overfit the data.

    “These values are consistent with the cross-validation scores and gave a similar receiver operating characteristic (ROC) curve (Fig. 3e), indicating that the model did not overfit the data.”
    Quote in the paperSubstantial claim

  • Classified as probably not a substantial claim.

    Study design and conductown data

    For robustness, the study investigated five standard machine learning algorithms with nested levels of optimization processes: model hyperparameters, model choice and multilevel validation.

    “For robustness, we investigated five standard machine learning algorithms with nested levels of optimization processes: model hyperparameters, model choice and multilevel validation.”
    Quote in the paperSubstantial claim

  • Classified as probably not a substantial claim.

    Premisecitation

    Organic colour centres (OCCs) are molecularly tunable quantum defects on SWCNTs produced by covalent functionalization.

    “Organic colour centres (OCCs) are molecularly tunable quantum defects on SWCNTs, which are produced by covalent functionalization of SWCNTs30,31.”
    Quote in the paperSubstantial claim0 of 2 sources checked
    • cites [30] Brozena, A. H., Kim, M., Powell, L. R. & Wang, Y. Controlling the optical properties of carbon nanotubes with organic colour-centre quantum defects. Nat. Rev. Chem. 3, 375–392 (2019). source

      Not checked in the free run.

    • cites [31] Piao, Y. M. et al. Brightening of carbon nanotube photoluminescence through the incorporation of sp3 defects. Nat. Chem. 5, 840–845 (2013). source

      Not checked in the free run.

  • Classified as probably not a substantial claim.

    Study design and conductnone given

    The averaged F-score in 10-fold cross-validation was used to assess model performance.

    “The averaged F-score in 10-fold cross-validation was used to assess the model performance”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

    Study design and conductnone given

    Potential interferents, such as drug treatments, were accounted for.

    “Potential interferents, such as drug treatments, were accounted for.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Premisecitation

    OCC fluorescence is molecularly specific and extremely sensitive to the local chemical environment of the atomic defect sites.

    “OCC fluorescence, on the other hand, is molecularly specific and extremely sensitive to the local chemical environment of the atomic defect sites32,41.”
    Quote in the paperSubstantial claim0 of 2 sources checked
    • cites [32] Kwon, H. et al. Optical probing of local pH and temperature in complex fluids with covalently functionalized, semiconducting carbon nanotubes. J. Phys. Chem. C 119, 3733–3739 (2015). source

      Not checked in the free run.

    • cites [41] Wu, X., Kim, M., Qu, H. & Wang, Y. Single-defect spectroscopy in the shortwave infrared. Nat. Commun. 10, 2672 (2019). source

      Not checked in the free run.

  • May combine two claims.

    Premisecitation

    Non-covalent encapsulation of SWCNTs with polymers, including short oligonucleotides, facilitates aqueous suspension and confers molecular selectivity to their optical responses via a molecular masking effect and modulation of optical bandgaps.

    “Non-covalent encapsulation with polymers, including short oligonucleotides, facilitates aqueous suspension and confers molecular selectivity to their optical responses via (1) contributing to a molecular masking effect that defines the shape and size of the exposed surface of SWCNTs26-28 and (2) modulating their optical bandgaps29.”
    Quote in the paperSubstantial claim0 of 4 sources checked
    • cites [26] Roxbury, D., Mittal, J. & Jagota, A. Molecular-basis of single-walled carbon nanotube recognition by single-stranded DNA. Nano Lett. 12, 1464–1469 (2012). source

      Not checked in the free run.

    • cites [27] Roxbury, D., Jagota, A. & Mittal, J. Structural characteristics of oligomeric DNA strands adsorbed onto single-walled carbon nanotubes. J. Phys. Chem. B 117, 132–140 (2013). source

      Not checked in the free run.

    • cites [28] Roxbury, D., Tu, X., Zheng, M. & Jagota, A. Recognition ability of DNA for carbon nanotubes correlates with their binding affinity. Langmuir 27, 8282–8293 (2011). source

      Not checked in the free run.

    • cites [29] Horoszko, C. P., Jena, P. V., Roxbury, D., Rotkin, S. V. & Heller, D. A. Optical voltammetry of polymer-encapsulated single-walled carbon nanotubes. J. Phys. Chem. C 123, 24200–24208 (2019). source

      Not checked in the free run.

  • May combine two claims.

    Premisecitation

    OCCs efficiently harvest mobile excitons through the SWCNT antenna, producing distinct E11− fluorescence bands and introducing new biochemical sensitivities to SWCNTs.

    “OCCs efficiently harvest mobile excitons through the SWCNT antenna, producing distinct fluorescence bands (E11−) at longer wavelengths from the E11 band. The E11− fluorescence introduces new biochemical sensitivities to SWCNTs determined by the chemical nature of the defect, making OCCs the molecular focal points for local environmental responses32.”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites [32] Kwon, H. et al. Optical probing of local pH and temperature in complex fluids with covalently functionalized, semiconducting carbon nanotubes. J. Phys. Chem. C 119, 3733–3739 (2015). source

      Not checked in the free run.

  • May combine two claims.

    Interpretationreasoning

    OCC-DNAs exhibit relatively narrow bandwidths (35-80 meV), making small spectral shifts easier to resolve compared with conventional fluorophores (>100 meV).

    “This is likely because OCC-DNAs exhibit relatively narrow bandwidths (35–80 meV), making small spectral shifts much easier to resolve compared with conventional fluorophores (>100 meV).”
    Quote in the paperSubstantial claim

  • Classified as probably not a substantial claim. May combine two claims.

    Premisecitation

    Machine learning processes recognize disease-specific patterns that are too subtle or complex for human eyes or conventional analytical methods and aid in the construction of robust diagnostic models.

    “Machine learning processes recognize disease-specific patterns that are too subtle or complex to be detected by human eyes or conventional analytical methods, and aid in the construction of robust diagnostic models17,18.”
    Quote in the paperSubstantial claim0 of 2 sources checked
    • cites [17] Bera, K., Schalper, K. A., Rimm, D. L., Velcheti, V. & Madabhushi, A. Artificial intelligence in digital pathology—new tools for diagnosis and precision oncology. Nat. Rev. Clin. Oncol. 16, 703–715 (2019). source

      Not checked in the free run.

    • cites [18] Rajkomar, A. et al. Scalable and accurate deep learning with electronic health records. NPJ Digit. Med. 1, 18 (2018). source

      Not checked in the free run.

  • May combine two claims.

    Interpretationreasoning

    Prediction outcomes will be lower in the clinical screening setting in which specimens are obtained in asymptomatic individuals before clinical diagnosis, because specimens obtained from symptomatic individuals at diagnosis were used for development and assessment.

    “However, considering the fact that specimens obtained from symptomatic individuals at diagnosis were used for the development and assessment of the technology, prediction outcomes will be lower in the clinical screening setting in which specimens are obtained in asymptomatic individuals before clinical diagnosis.”
    Quote in the paperSubstantial claim
    • Rests on an assumption: Serum biomarker profiles differ between symptomatic individuals at diagnosis and asymptomatic individuals before clinical diagnosis.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Study design and conductcitation

    The risk of bias in the study was evaluated using PROBAST.

    “The risk of bias in the study was evaluated on the basis of Prediction model Risk Of Bias Assessment, PROBLAST36 (Appendix 1 in Supplementary Information).”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [36] Wolff, R. F. et al. PROBAST: a tool to assess the risk of bias and applicability of prediction model studies. Ann. Intern. Med. 170, 51–58 (2019). source

      Not checked in the free run.

  • May combine two claims.

    Premisecitation

    The emission of SWCNTs (E11) is sensitive to dielectric environments, redox perturbations and electrostatic charge.

    “The emission of SWCNTs (E11) is sensitive to dielectric environments21,22, redox perturbations23 and electrostatic charge24,25.”
    Quote in the paperSubstantial claim0 of 5 sources checked
    • cites [21] Heller, D. A. et al. Optical detection of DNA conformational polymorphism on single-walled carbon nanotubes. Science 311, 508 (2006). source

      Not checked in the free run.

    • cites [22] Jena, P. V. et al. A carbon nanotube optical reporter maps endolysosomal lipid flux. ACS Nano 11, 10689–10703 (2017). source

      Not checked in the free run.

    • cites [23] Heller, D. A. et al. Multimodal optical sensing and analyte specificity using single-walled carbon nanotubes. Nat. Nanotech. 4, 114–120 (2009). source

      Not checked in the free run.

    • cites [24] Roxbury, D., Jena, P. V., Shamay, Y., Horoszko, C. P. & Heller, D. A. Cell membrane proteins modulate the carbon nanotube optical bandgap via surface charge accumulation. ACS Nano 10, 499–506 (2016). source

      Not checked in the free run.

    • cites [25] Williams, R. M. et al. Noninvasive ovarian cancer biomarker detection via an optical nanosensor implant. Sci. Adv. 4, eaaq1090 (2018). source

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Study design and conductnone given

    There was no statistically significant difference in age distribution for each group.

    “There was no statistically significant difference in age distribution for each group.”
    Quote in the paperNot a substantial claim

  • Classified as probably not a substantial claim.

    Study design and conductcitation

    The imbalance in the size of each group was corrected by upscaling minority species (SMOTE) so that the prediction models were not biased by groups with larger sample sizes.

    “The imbalance in the size of each group was corrected by upscaling minority species (SMOTE: Synthetic Minority Oversampling Technique)49 so that the prediction models were not biased by groups with larger sample sizes.”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites [49] Chawla, N. V., Bowyer, K. W., Hall, L. O. & Kegelmeyer, W. P. SMOTE: synthetic minority over-sampling technique. J. Artif. Intell. Res. 16, 321–357 (2002). source

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Study design and conductnone given

    All experiments were performed in triplicate.

    “All experiments were performed in triplicate.”
    Quote in the paperNot a substantial claim

  • Classified as probably not a substantial claim.

    Study design and conductreasoning

    Model performance was evaluated using 10-fold cross-validation to rule out possible overfitting in the machine learning process.

    “To rule out possible overfitting in the machine learning process, model performance was evaluated using 10-fold cross-validation.”
    Quote in the paperSubstantial claim

  • Classified as probably not a substantial claim. May combine two claims.

    Study design and conductnone given

    For the biomarker prediction models, the data were divided into normal versus high biomarker level groups on the basis of clinical references (CA125, 50 U ml−1; HE4, 150 pM; YKL40, 1,650 pM) and the group size was corrected using SMOTE.

    “For the biomarker prediction models, we divided the data into normal versus high biomarker level groups on the basis of clinical references (CA125, 50 U ml−1; HE4, 150 pM; YKL40, 1,650 pM) and corrected the group size using SMOTE.”
    Quote in the paperSubstantial claim

  • Classified as probably not a substantial claim. May combine two claims.

    Study design and conductnone given

    Diagnoses were identified from a chart review of each patient; all diagnoses included histology and were confirmed by a gynaecologic oncology attending physician.

    “Diagnoses were identified from a chart review of each patient; all diagnoses included histology and were confirmed by a gynaecologic oncology attending physician.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Study design and conductnone given

    Each sample was measured in triplicate.

    “Each sample was measured in triplicate.”
    Quote in the paperNot a substantial claim

  • Classified as probably not a substantial claim.

    Study design and conductown data

    The trained models were tested with an independent set of patient sera (N = 54), sampled from different patients, as external validation.

    “The trained models were then tested with an independent set of patient sera (N = 54), sampled from different patients (test set), as external validation.”
    Quote in the paperSubstantial claim

  • Classified as probably not a substantial claim.

    Study design and conductown data

    Six OCC-DNA nanosensors, with p-values of the spectroscopic features lower than 0.10, were selected for the sensor array.

    “Six OCC-DNA nanosensors, with p-values of the spectroscopic features lower than 0.10, were selected for the sensor array.”
    Quote in the paperSubstantial claim
41 of 41

  • The paper relies on this without support.

    The current best clinical screening test uses CA125 and transvaginal ultrasonography and has 84% sensitivity at 98% specificity.

    “compared with 84% sensitivity at 98% specificity for the current best clinical screening test, which uses measurements of cancer antigen 125 and transvaginal ultrasonography”

  • The paper relies on this without support.

    Earlier detection of ovarian cancer and timely measurements of disease progression and recurrence would markedly improve outcomes.

    “The earlier detection of ovarian cancer and timely measurements of disease progression and recurrence would markedly improve outcomes.”

  • The paper relies on this without support.

    Accurate detection of known analytes does not always confer high sensitivity and specificity for disease.

    “However, accurate detection of known analytes does not always confer high sensitivity and specificity for disease.”

  • The paper relies on this without support.

    Serum biomarker profiles differ between symptomatic individuals at diagnosis and asymptomatic individuals before clinical diagnosis.

    “However, considering the fact that specimens obtained from symptomatic individuals at diagnosis were used for the development and assessment of the technology, prediction outcomes will be lower in the clinical screening setting in which specimens are obtained in asymptomatic individuals before clinical diagnosis.”

  • The paper relies on this without support.

    The spectroscopic variables that are not sensitive to known biomarkers reflect the presence of unknown biomarkers or combinations thereof.

    “This suggests that there exist potential biomarkers or combinations thereof that are either unknown or not part of conventional screening approaches but were captured by the OCC-DNA sensor array.”
5 of 5

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

  • A nanosensor array technology combined with machine learning algorithms can distinguish between ovarian cancer patients and healthy individuals and predict ovarian cancer with high sensitivity and specificity.

    “Kim et al. constructed a nanosensor array technology to identify disease spectral fingerprints and combined it with machine learning algorithms to successfully distinguish between ovarian cancer patients and healthy individuals, which can be used to predict ovarian cancer with high sensitivity and specificity.344”

    From Smart nanoparticles for cancer therapy (2023) · cites this paper as [344]

The citation lists haven’t been fetched yet

References 51 · Cited by 214

Research works printed but not linked: 1

  1. Pedregosa, F. et al. Scikit-learn: machine learning in Python. J. Mach. Learn. Res. 12, 2825–2830 (2011).

Authors
Kim, M., Chen, C., Wang, P., Mulvey, J. J., Yang, Y., Wun, C., Antman-Passig, M., Luo, H.-B., Cho, S., Long-Roche, K., Ramanathan, L. V., Jagota, A., Zheng, M., Wang, Y., & Heller, D. A.
Affiliations
  • Memorial Sloan Kettering Cancer Center (US)
  • University of Maryland, College Park (US)
  • Albert Einstein College of Medicine (US)
  • Lehigh University (US)
  • Hunter College (US)
  • National Institute of Standards and Technology (US)
Corresponding author
Heller, Daniel A.
ORCID
Venue
Nature Biomedical Engineering
Publisher
Springer Science and Business Media LLC
Published
2022-03-17
Volume
6
Issue
3
Pages
267-275
ISSN
2157-846X (ISSN-L)
DOI
10.1038/s41551-022-00860-y
PMC
PMC9108893
OpenAlex
W4221014558
Cited by
214
References
51
Citations by year
2026: 45 · 2025: 61 · 2024: 56 · 2023: 36 · 2022: 16
FWCI
18.75
Type
journal article
Language
en
Open access
Licence
Locations
  • Nature Biomedical Engineering (publishedVersion)
  • PubMed (publishedVersion)
  • PubMed Central (acceptedVersion)
Funders
  • National Institutes of Health (5r01ca215719-05)
  • National Cancer Institute (R01-CA215719, P30CA008748, U54CA137788, U54CA132378)
  • National Institute of General Medical Sciences (R01GM114167)
Topics
  • Life Sciences › Biochemistry, Genetics and Molecular Biology › Biophysics › Spectroscopy Techniques in Biomedical and Chemical Research
  • Health Sciences › Medicine › Reproductive Medicine › Ovarian cancer diagnosis and treatment
  • Life Sciences › Biochemistry, Genetics and Molecular Biology › Molecular Biology › Advanced biosensing and bioanalysis techniques
Keywords
ovarian cancer, high-grade serous ovarian cancer, near-infrared fluorescence, spectral fingerprinting, serum biomarkers, cancer screening, machine learning classification, early cancer detection
MeSH terms
Machine Learning, Female, Humans, Ovarian Neoplasms, Biomarkers, Tumor, Nanotubes, Carbon, Early Detection of Cancer

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Claims read by deepseek-v4.1-flash under definition version 3 (extract-v3) on 2026-10-10.

Classified by jev-1.13-20260917 (gold-guideline-v0.2) on 2026-10-10.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review