Harnessing Consumer Wearable Digital Biomarkers for Individualized Recognition of Postpartum Depression Using the All of Us Research Program Data Set: Cross-Sectional Study.

BACKGROUND: Postpartum depression (PPD) poses a significant maternal health challenge. The current approach to detecting PPD relies on in-person postpartum visits, which contributes to underdiagnosis. Furthermore, recognizing PPD symptoms can be challenging. Therefore, we explored the potential of using digital biomarkers from consumer wearables for PPD recognition. OBJECTIVE: The main goal of this study was to showcase the viability of using machine learning (ML) and digital biomarkers related to heart rate, physical activity, and energy expenditure derived from consumer-grade wearables for the recognition of PPD. METHODS: Using the All of Us Research Program Registered Tier v6 data set, we performed computational phenotyping of women with and without PPD following childbirth. Intraindividual ML models were developed using digital biomarkers from Fitbit to discern between prepregnancy, pregnancy, postpartum without depression, and postpartum with depression (ie, PPD diagnosis) periods. Models were built using generalized linear models, random forest, support vector machine, and k-nearest neighbor algorithms and evaluated using the κ statistic and multiclass area under the receiver operating characteristic curve (mAUC) to determine the algorithm with the best performance. The specificity of our individualized ML approach was confirmed in a cohort of women who gave birth and did not experience PPD. Moreover, we assessed the impact of a previous history of depression on model performance. We determined the variable importance for predicting the PPD period using Shapley additive explanations and confirmed the results using a permutation approach. Finally, we compared our individualized ML methodology against a traditional cohort-based ML model for PPD recognition and compared model performance using sensitivity, specificity, precision, recall, and F1-score. RESULTS: Patient cohorts of women with valid Fitbit data who gave birth included <20 with PPD and 39 without PPD. Our results demonstrated that intraindividual models using digital biomarkers discerned among prepregnancy, pregnancy, postpartum without depression, and postpartum with depression (ie, PPD diagnosis) periods, with random forest (mAUC=0.85; κ=0.80) models outperforming generalized linear models (mAUC=0.82; κ=0.74), support vector machine (mAUC=0.75; κ=0.72), and k-nearest neighbor (mAUC=0.74; κ=0.62). Model performance decreased in women without PPD, illustrating the method's specificity. Previous depression history did not impact the efficacy of the model for PPD recognition. Moreover, we found that the most predictive biomarker of PPD was calories burned during the basal metabolic rate. Finally, individualized models surpassed the performance of a conventional cohort-based model for PPD detection. CONCLUSIONS: This research establishes consumer wearables as a promising tool for PPD identification and highlights personalized ML approaches, which could transform early disease detection strategies.

Session Introduction: Digital health technology data in biocomputing: Research efforts and considerations for expanding access (PSB2024).

Data from digital health technologies (DHT), including wearable sensors like Apple Watch, Whoop, Oura Ring, and Fitbit, are increasingly being used in biomedical research. Research and development of DHT-related devices, platforms, and applications is happening rapidly and with significant private-sector involvement with new biotech companies and large tech companies (e.g. Google, Apple, Amazon, Uber) investing heavily in technologies to improve human health. Many academic institutions are building capabilities related to DHT research, often in cross-sector collaboration with technology companies and other organizations with the goal of generating clinically meaningful evidence to improve patient care, to identify users at an earlier stage of disease presentation, and to support health preservation and disease prevention. Large research consortia, cross-sector partnerships, and individual research labs are all represented in the current corpus of published studies. Some of the large research studies, like NIH's All of Us Research Program, make data sets from wearable sensors available to the research community, while the vast majority of data from wearable sensors and other DHTs are held by private sector organizations and are not readily available to the research community. As data are unlocked from the private sector and made available to the academic research community, there is an opportunity to develop innovative analytics and methods through expanded access. This is the second year for this Session which solicited research results leveraging digital health technologies, including wearable sensor data, describing novel analytical methods, and issues related to diversity, equity, inclusion (DEI) of the research, data, and the community of researchers working in this area. We particularly encouraged submissions describing opportunities for expanding and democratizing academic research using data from wearable sensors and related digital health technologies.

Harnessing consumer wearable digital biomarkers for individualized recognition of postpartum depression using the All of Us Research Program dataset.

Postpartum depression (PPD), afflicting one in seven women, poses a major challenge in maternal health. Existing approaches to detect PPD heavily depend on in-person postpartum visits, leading to cases of the condition being overlooked and untreated. We explored the potential of consumer wearable-derived digital biomarkers for PPD recognition to address this gap. Our study demonstrated that intra-individual machine learning (ML) models developed using these digital biomarkers can discern between pre-pregnancy, pregnancy, postpartum without depression, and postpartum with depression time periods (i.e., PPD diagnosis). When evaluating variable importance, calories burned from the basal metabolic rate (calories BMR) emerged as the digital biomarker most predictive of PPD. To confirm the specificity of our method, we demonstrated that models developed in women without PPD could not accurately classify the PPD-equivalent phase. Prior depression history did not alter model efficacy for PPD recognition. Furthermore, the individualized models demonstrated superior performance compared to a conventional cohort-based model for the detection of PPD, underscoring the effectiveness of our individualized ML approach. This work establishes consumer wearables as a promising avenue for PPD identification. More importantly, it also emphasizes the utility of individualized ML model methodology, potentially transforming early disease detection strategies.

Session Introduction: Digital health technology data in biocomputing: Research efforts and considerations for expanding access.

Data from digital health technologies (DHT), including wearable sensors like Apple Watch, Whoop, Oura Ring, and Fitbit, are increasingly being used in biomedical research. Research and development of DHT-related devices, platforms, and applications is happening rapidly and with significant private-sector involvement with new biotech companies and large tech companies (e.g. Google, Apple, Amazon, Uber) investing heavily in technologies to improve human health. Many academic institutions are building capabilities related to DHT research, often in cross-sector collaboration with technology companies and other organizations with the goal of generating clinically meaningful evidence to improve patient care, to identify users at an earlier stage of disease presentation, and to support health preservation and disease prevention. Large research consortia, cross-sector partnerships, and individual research labs are all represented in the current corpus of published studies. Some of the large research studies, like NIH's All of Us Research Program, make data sets from wearable sensors available to the research community, while the vast majority of data from wearable sensors and other DHTs are held by private sector organizations and are not readily available to the research community. As data are unlocked from the private sector and made available to the academic research community, there is an opportunity to develop innovative analytics and methods through expanded access. This Session solicited research results leveraging digital health technologies, including wearable sensor data, describing novel analytical methods, and issues related to diversity, equity, inclusion (DEI) of both the underlying research data sets and the community of researchers working in this area. We particularly encouraged submissions describing opportunities for expanding and democratizing academic research using data from wearable sensors and related digital health technologies.

How Fitbit data are being made available to registered researchers in All of Us Research Program.

The National Institutes of Health's (NIH) All of Us Research Program aims to enroll at least one million US participants from diverse backgrounds; collect electronic health record (EHR) data, survey data, physical measurements, biospecimens for genomics and other assays, and digital health data; and create a researcher database and tools to enable precision medicine research [1]. Since inception, digital health technologies (DHT) have been envisioned as essential to achieving the goals of the program [2]. A "bring your own device" (BYOD) study for collecting Fitbit data from participants' devices was developed with integration of additional DHTs planned in the future [3]. Here we describe how participants can consent to share their digital health technology data, how the data are collected, how the data set is parsed, and how researchers can access the data.

Biomedical research in the Cloud: considerations for researchers and organizations moving to (or adding) cloud computing resources.

As biomedical research data grow, researchers need reliable and scalable solutions for storage and compute. There is also a need to build systems that encourage and support collaboration and data sharing, to result in greater reproducibility. This has led many researchers and organizations to use cloud computing [1]. The cloud not only enables scalable, on-demand resources for storage and compute, but also collaboration and continuity during virtual work, and can provide superior security and compliance features. Moving to or adding cloud resources, however, is not trivial or without cost, and may not be the best choice in every scenario. The goal of this workshop is to explore the benefits of using the cloud in biomedical and computational research, and considerations (pros and cons) for a range of scenarios including individual researchers, collaborative research teams, consortia research programs, and large biomedical research agencies / organizations.

Wearable fitness tracker use in federally qualified health center patients: strategies to improve the health of all of us using digital health devices.

As the use of connected devices rises, an understanding of how digital health technologies can be used for equitable healthcare across diverse communities is needed. We surveyed 1007 adult patients at six Federally Qualified Health Centers regarding wearable fitness trackers. Findings indicate the majority interest in having fitness trackers. Barriers included cost and lack of information, revealing that broad digital health device adoption requires education, investment, and high-touch methods.

Lipid metabolism genes in contralateral unaffected breast and estrogen receptor status of breast cancer.

Risk biomarkers that are specific to estrogen receptor (ER) subtypes of breast cancer would aid the development and implementation of distinct prevention strategies. The contralateral unaffected breast of women with unilateral breast cancer (cases) is a good model for defining subtype-specific risk because women with ER-negative (ER-) index primaries are at high risk for subsequent ER-negative primary cancers. We conducted random fine needle aspiration of the unaffected breasts of cases. Samples from 30 subjects [15 ER-positive (ER+) and 15 ER- cases matched for age, race and menopausal status] were used for Illumina expression array analysis. Findings were confirmed using quantitative real-time PCR (qRT-PCR) in the same samples. A validation set consisting of 36 subjects (12 ER+, 12 ER- and 12 standard-risk healthy controls) was used to compare gene expression across groups. ER- case samples displayed significantly higher expression of 18 genes/transcripts, 8 of which were associated with lipid metabolism on gene ontology analysis (GO: 0006629). This pattern was confirmed by qRT-PCR in the same samples, and in the 24 cases of the validation set. When compared to the healthy controls in the validation set, significant overexpression of 4 genes (DHRS2, HMGCS2, HPGD and ACSL3) was observed in ER- cases, with significantly lower expression of UGT2B11 and APOD in ER+ cases, and decreased expression of UGT2B7 in both subtypes. These data suggest that differential expression of lipid metabolism genes may be involved in the risk for subtypes of breast cancer, and are potential biomarkers of ER-specific breast cancer risk.

NCBI GEO: archive for functional genomics data sets--update.

The Gene Expression Omnibus (GEO, http://www.ncbi.nlm.nih.gov/geo/) is an international public repository for high-throughput microarray and next-generation sequence functional genomic data sets submitted by the research community. The resource supports archiving of raw data, processed data and metadata which are indexed, cross-linked and searchable. All data are freely available for download in a variety of formats. GEO also provides several web-based tools and strategies to assist users to query, analyse and visualize data. This article reports current status and recent database developments, including the release of GEO2R, an R-based web application that helps users analyse GEO data.

NCBI GEO: archive for functional genomics data sets--10 years on.

A decade ago, the Gene Expression Omnibus (GEO) database was established at the National Center for Biotechnology Information (NCBI). The original objective of GEO was to serve as a public repository for high-throughput gene expression data generated mostly by microarray technology. However, the research community quickly applied microarrays to non-gene-expression studies, including examination of genome copy number variation and genome-wide profiling of DNA-binding proteins. Because the GEO database was designed with a flexible structure, it was possible to quickly adapt the repository to store these data types. More recently, as the microarray community switches to next-generation sequencing technologies, GEO has again adapted to host these data sets. Today, GEO stores over 20,000 microarray- and sequence-based functional genomics studies, and continues to handle the majority of direct high-throughput data submissions from the research community. Multiple mechanisms are provided to help users effectively search, browse, download and visualize the data at the level of individual genes or entire studies. This paper describes recent database enhancements, including new search and data representation tools, as well as a brief review of how the community uses GEO data. GEO is freely accessible at http://www.ncbi.nlm.nih.gov/geo/.

Annotating the human genome with Disease Ontology.

BACKGROUND: The human genome has been extensively annotated with Gene Ontology for biological functions, but minimally computationally annotated for diseases. RESULTS: We used the Unified Medical Language System (UMLS) MetaMap Transfer tool (MMTx) to discover gene-disease relationships from the GeneRIF database. We utilized a comprehensive subset of UMLS, which is disease-focused and structured as a directed acyclic graph (the Disease Ontology), to filter and interpret results from MMTx. The results were validated against the Homayouni gene collection using recall and precision measurements. We compared our results with the widely used Online Mendelian Inheritance in Man (OMIM) annotations. CONCLUSION: The validation data set suggests a 91% recall rate and 97% precision rate of disease annotation using GeneRIF, in contrast with a 22% recall and 98% precision using OMIM. Our thesaurus-based approach allows for comparisons to be made between disease containing databases and allows for increased accuracy in disease identification through synonym matching. The much higher recall rate of our approach demonstrates that annotating human genome with Disease Ontology and GeneRIF for diseases dramatically increases the coverage of the disease annotation of human genome.

Promyelocytic leukemia zinc finger protein regulates interferon-mediated innate immunity.

Interferons (IFNs) direct innate and acquired immune responses and, accordingly, are used therapeutically to treat a number of diseases, yet the diverse effects they elicit are not fully understood. Here, we identified the promyelocytic leukemia zinc finger (PLZF) protein as a previously unrecognized component of the IFN response. IFN stimulated an association of PLZF with promyelocytic leukemia protein (PML) and histone deacetylase 1 (HDAC1) to induce a decisive subset of IFN-stimulated genes (ISGs). Consequently, PLZF-deficient mice had a specific ISG expression defect and as a result were more susceptible to viral infection. This susceptibility correlated with a marked decrease in the expression of the key antiviral mediators and an impaired IFN-mediated induction of natural killer cell function. These results provide new insights into the regulatory mechanisms of IFN signaling and the induction of innate antiviral immunity.

From disease ontology to disease-ontology lite: statistical methods to adapt a general-purpose ontology for the test of gene-ontology associations.

Subjective methods have been reported to adapt a general-purpose ontology for a specific application. For example, Gene Ontology (GO) Slim was created from GO to generate a highly aggregated report of the human-genome annotation. We propose statistical methods to adapt the general purpose, OBO Foundry Disease Ontology (DO) for the identification of gene-disease associations. Thus, we need a simplified definition of disease categories derived from implicated genes. On the basis of the assumption that the DO terms having similar associated genes are closely related, we group the DO terms based on the similarity of gene-to-DO mapping profiles. Two types of binary distance metrics are defined to measure the overall and subset similarity between DO terms. A compactness-scalable fuzzy clustering method is then applied to group similar DO terms. To reduce false clustering, the semantic similarities between DO terms are also used to constrain clustering results. As such, the DO terms are aggregated and the redundant DO terms are largely removed. Using these methods, we constructed a simplified vocabulary list from the DO called Disease Ontology Lite (DOLite). We demonstrated that DOLite results in more interpretable results than DO for gene-disease association tests. The resultant DOLite has been used in the Functional Disease Ontology (FunDO) Web application at http://www.projects.bioinformatics.northwestern.edu/fundo.

Functional annotation of IFN-alpha-stimulated gene expression profiles from sensitive and resistant renal cell carcinoma cell lines.

The antiproliferative, antiviral, and immunomodulatory properties of interferons (IFNs) have led to its therapeutic implementation. IFNs effects are mediated by a complex network of signal transducers, culminating in IFN-stimulated gene (ISG) induction. This complexity leads to diverse clinical responses to IFN, from no response to complete regression of disease. Elucidation of ISG induction patterns is, therefore, essential to understand and maximize its therapeutic potential. To correlate ISG expression profiles with IFN responsiveness, two renal cell carcinoma (RCC) cell lines differing in antiviral and apoptotic response to IFN were treated with IFN-alpha for different times, and expression profiles were analyzed using a customized microarray containing 850 unique putative ISGs. Genes with similar kinetics of induction in both cell lines were clustered and analyzed for gene function. Seven sets of coordinately regulated genes were identified by k-means cluster analysis, and significant functional similarities were identified for five of the seven sets. Strikingly, expression of genes associated with transcription temporally preceded expression of those involved in signal transduction. Enhanced antiviral sensitivity to IFN was coincident with sustained expression of ISGs involved in transcriptional regulation. However, no difference in Stat1 activation was observed between the cell lines. Analysis of ISG expression patterns suggests that subtle differences in transcription profiles contribute to differences in IFN responsiveness.

Expression of IFITM1 in chronic myeloid leukemia patients.

We investigated the peripheral blood gene expression profile of interferon induced transmembrane protein 1 (IFITM1) in sixty chronic myeloid leukemia (CML) patients classified according to new prognostic score (NPS). IFITM1 is a component of a multimeric complex involved in the trunsduction of antiproliferative and cell adhesion signals. Expression level of IFITM1 was found significantly different between the high- and low-risk groups (P = 9.7976 x 10(-11)) by real-time reverse transcription polymerase chain reaction (RT-PCR). Higher IFITM1 expression correlated with improved survival (P = 0.01). These results indicate that IFITM1 expression profiling could be used for molecular classification of CML, which may also predict survival.

Induction of interferon-stimulated gene expression and antiviral responses require protein deacetylase activity.

Histone deacetylase (HDAC) activity, commonly correlated with transcriptional repression, was essential for transcriptional induction of IFN-stimulated genes (ISG). Inhibition of HDAC function led to global impairment of ISG expression, with little effect on basal expression. HDAC function was not required for signal transducer and activator of transcription tyrosine phosphorylation, nuclear translocation, or assembly on chromatin, but it was needed for full activity of the signal transducer and activator of transcription transactivation domain. HDAC function was also required for gene induction driven by the IFN regulatory factor 3 transcription factor activated by virus infection, and it was essential for establishment of an antiviral response against Flaviviridae, Rhabdoviridae, and Picornaviridae. Requirement for HDAC function in transcriptional activation may represent a general mechanism for rapid stimulation of ISG transcription.

Activation of the interferon system by short-interfering RNAs.

RNA interference (RNAi) is a powerful tool used to manipulate gene expression or determine gene function. One technique of expressing the short double-stranded (ds) RNA intermediates required for interference in mammalian systems is the introduction of short-interfering (si) RNAs. Although RNAi strategies are reliant on a high degree of specificity, little attention has been given to the potential non-specific effects that might be induced. Here, we found that transfection of siRNAs results in interferon (IFN)-mediated activation of the Jak-Stat pathway and global upregulation of IFN-stimulated genes. This effect is mediated by the dsRNA-dependent protein kinase, PKR, which is activated by 21-base-pair (bp) siRNAs and required for upregulation of IFN-beta in response to siRNAs. In addition, we show by using cell lines deficient in specific components mediating IFN action that the RNAi mechanism itself is independent of the interferon system. Thus, siRNAs have broad and complicating effects beyond the selective silencing of target genes when introduced into cells. This is of critical importance, as siRNAs are currently being explored for their potential therapeutic use.