Project description:BackgroundPredicting the likelihood of success of weight loss interventions using machine learning (ML) models may enhance intervention effectiveness by enabling timely and dynamic modification of intervention components for nonresponders to treatment. However, a lack of understanding and trust in these ML models impacts adoption among weight management experts. Recent advances in the field of explainable artificial intelligence enable the interpretation of ML models, yet it is unknown whether they enhance model understanding, trust, and adoption among weight management experts.ObjectiveThis study aimed to build and evaluate an ML model that can predict 6-month weight loss success (ie, ≥7% weight loss) from 5 engagement and diet-related features collected over the initial 2 weeks of an intervention, to assess whether providing ML-based explanations increases weight management experts' agreement with ML model predictions, and to inform factors that influence the understanding and trust of ML models to advance explainability in early prediction of weight loss among weight management experts.MethodsWe trained an ML model using the random forest (RF) algorithm and data from a 6-month weight loss intervention (N=419). We leveraged findings from existing explainability metrics to develop Prime Implicant Maintenance of Outcome (PRIMO), an interactive tool to understand predictions made by the RF model. We asked 14 weight management experts to predict hypothetical participants' weight loss success before and after using PRIMO. We compared PRIMO with 2 other explainability methods, one based on feature ranking and the other based on conditional probability. We used generalized linear mixed-effects models to evaluate participants' agreement with ML predictions and conducted likelihood ratio tests to examine the relationship between explainability methods and outcomes for nested models. We conducted guided interviews and thematic analysis to study the impact of our tool on experts' understanding and trust in the model.ResultsOur RF model had 81% accuracy in the early prediction of weight loss success. Weight management experts were significantly more likely to agree with the model when using PRIMO (χ2=7.9; P=.02) compared with the other 2 methods with odds ratios of 2.52 (95% CI 0.91-7.69) and 3.95 (95% CI 1.50-11.76). From our study, we inferred that our software not only influenced experts' understanding and trust but also impacted decision-making. Several themes were identified through interviews: preference for multiple explanation types, need to visualize uncertainty in explanations provided by PRIMO, and need for model performance metrics on similar participant test instances.ConclusionsOur results show the potential for weight management experts to agree with the ML-based early prediction of success in weight loss treatment programs, enabling timely and dynamic modification of intervention components to enhance intervention effectiveness. Our findings provide methods for advancing the understandability and trust of ML models among weight management experts.

Project description:Pluripotency can be induced from different cell types using different methods. Several studies have reported that iPSC retain an epigenetic memory from the tissue from which they were derived. However, the influence of the method used for reprogramming in the final pluripotent state has not been investigated yet. Therefore, we have taken advantage of germline stem cells (GSCs) that can be converted into germline pluripotent stem cells (gPSCs) using specific culture conditions or into induced pluripotent stem cells (iPSCs) after the forced expression of four transcription factors. Here, we compare the gene expression profile, imprinting methylation pattern and differentiation potential between GSC-derived iPSCs and gPSCs. Our analysis demonstrates that both reprogramming methods induce equivalent levels of pluripotency and that a donor GSC transcriptional memory is retained only at early steps after reprogramming. In summary, our results show that different reprogramming methods generate undistinguishable pluripotent states and exclude a method-dependent epigenetic memory. Cell Culture. All the cell lines used in this study were derived from OG2 mice. Unless otherwise noted, all cell culture reagents were purchased from Gibco or PAA. Neonatal GSCs derived from 10 days postpartum (d.p.p) murine testis and adult GSCs derived from 35 d.p.p murine testis were derived as previously described and cultured on irradiated C3H feeder cells in a medium composed of StemPro-34 SFM (including supplement), N2 supplement (1:100), 1Â % heat-inactivated fetal calf serum (FCS), 5Â mgÂ ml-1 bovine serum albumin (BSA) fractionÂ V, 100Â UÂ ml-1 penicillin, 100Â ÂµgÂ ml-1 streptomycin, 2Â mM Lâglutamine, 0.1Â mM non-essential amino acids (NEAA), 0.1Â mM Î²âmercaptoethanol, MEM vitamins (1:100), 6Â mgÂ ml-1 Dâ(+)âglucose (Sigma), 30Â mgÂ ml-1 pyruvic acid, 0.1Â % DLâlactic acid (Sigma), 60Â ngÂ ml-1 progesterone (Sigma), 30Â ngÂ ml-1 Î²âestradiol (Sigma), 15Â ngÂ ml-1 mouse EGF (Peprotech), 10Â ngÂ ml-1 human bFGF (Peprotech), 10Â ngÂ ml-1 human GDNF (Peprotech), and 1000Â UÂ ml-1 mouse LIF (inâhouse preparation). For maintenance culture, GSCs were passaged at a 1:2âratio once a week after digestion with 0.05Â % trypsin/EDTA. In addition, ESCs, iPSC and gPSC were cultured on irradiated C3H feeders in the following medium: Knockout DMEM, 20% Knockout Serum Replacement, penicillin, streptomycin, Lâglutamine, NEAA, Î²âmercaptoethanol, and LIF. Generation of pluripotent cell lines. iPSC were generated using tetracycline-inducible lentiviruses. For the lentiviral production, 293T cells were plated at 50 % confluency in 10-cm dishes one day prior to transfection. The next day, the lentiviral vector (3 Âµg) together with the packaging plasmids psPAX2 (2Â Âµg) and pMD2.G (1Â Âµg) (Addgene plasmids #12260 and #12259, respectively) were transfected using FuGENE 6 reagent (Roche) according to the manufacturerâs protocol. After 48 h, viral-containing supernatants were filtered (0.45Â Âµm), concentrated around 30-fold by centrifugation (26,000 Ã g, 2Â h, 4Â Â°C) and stored at -80Â Â°C. For the generation of iPSCs, a single-cell suspension of 50,000 GSCs was infected overnight with five tetracycline-inducible lentiviruses (5â20Â Âµl concentrated supernatant each) encoding for Oct4, Sox2, Klf4, c-Myc, and the reverse tetracycline-transactivator M2-rtTA (Addgene plasmids #20323, #20326, #20322, #20324, and #20342) in the presence of 4Â ÂµgÂ ml-1 polybrene (Sigma). One week after the infection, the expression of the viral transgenes was induced by adding 4Â ÂµgÂ ml-1 doxycycline (Sigma) to the medium. MEFs-derived iPSCs were generated using the same protocol. gPSC were generated following a previous reported protocol. Briefly, GSCs were seeded at a density of approx. 1000 cells cm-2 on feeder cells and were not splitted until gPSC colonies were observed. Both iPSC and gPSC colonies were identified by Oct4-GFP expression, manually picked and expanded under ESC conditions. In vitro differentiation of pluripotent cells. Embryoid body (EB) formation was induced in suspension culture adopting the hanging drop method (600 cells in 20 Âµl) for 5 days in ESC medium without LIF. EBs were seeded on gelatine-coated plates and further cultured for 14 days in the same medium. Chimera formation. Aggregations of putative pluripotent cells with denuded 8-cell stage embryos were performed as previously reported. After overnight incubation at 37 Â°C and 5 % CO2, 11â14 aggregates were transferred into the uterine horns of 2.5 d.p.c pseudo-pregnant surrogate mothers (mated with vasectomized males). The chimeric pups were obtained and dissected at 14.5 d.p.c. Microarray gene expression analysis. Purified cRNA samples were prepared with the linear TotalPrep RNA Amplification Kit (Ambion) from 500 ng original RNA, involving synthesis of double-stranded cDNA by T7-promoter-linked oligo(dT) primers and 14 hours of in-vitro transcription incorporating biotin-labelled nucleotides. Hybridizations onto MouseRef 8 v2.0 Expression BeadChips (Illumina) was performed according to the manufacturerâs recommendations. Two replicates were hybridized from each sample. Chips were stained with streptavidin-Cy3 (GE Healthcare) and scanned using an iScan reader (Illumina). The bead intensities were mapped to gene information using BeadStudio 3.2 (Illumina). Background correction was performed using the Affymetrix robust multiarray analysis (RMA) background correction model. Variance stabilization was performed by log2 scaling, and gene expression normalization was calculated with the method implemented in the lumi package of R-Bioconductor. Samples were hybridised as biological replicates. 26 samples were analyzed. iPSC[GSCneonatal]#1: iPSC day 10 from GSC, clone 1, 2 biological rep iPSC[GSCneonatal]#2: iPSC day 10 from GSC, clone 2, 2 biological rep iPSC[GSCadult]#1: iPSC day 40 from GSC, clone 1, 2 biological rep iPSC[GSCadult]#2: iPSC day 40 from GSC, clone 2, 2 biological rep gPSC[GSCneonatal]#1: gPSC day 10 from GSC, clone 1, 2 biological rep gPSC[GSCneonatal]#2: gPSC day 10 from GSC, clone 2, 2 biological rep gPSC[GSCadult]#1: gPSC day 40 from GSC, clone 1, 2 biological rep gPSC[GSCadult]#2: gPSC day 40 from GSC, clone 2, 2 biological rep ESC: OG2 ESCs, 2 biological rep iPSC[MEF]: iPSC (Tet-4F) from OG2 MEFs, 2 biological rep GSC[neonatal]: GSC day 10, 2 biological rep GSC[adult]: GSC day 40, 2 biological rep MEF: Mouse embryonic fibroblasts, 2 biological rep

Project description:Pluripotency can be induced from different cell types using different methods. Several studies have reported that iPSC retain an epigenetic memory from the tissue from which they were derived. However, the influence of the method used for reprogramming in the final pluripotent state has not been investigated yet. Therefore, we have taken advantage of germline stem cells (GSCs) that can be converted into germline pluripotent stem cells (gPSCs) using specific culture conditions or into induced pluripotent stem cells (iPSCs) after the forced expression of four transcription factors. Here, we compare the gene expression profile, imprinting methylation pattern and differentiation potential between GSC-derived iPSCs and gPSCs. Our analysis demonstrates that both reprogramming methods induce equivalent levels of pluripotency and that a donor GSC transcriptional memory is retained only at early steps after reprogramming. In summary, our results show that different reprogramming methods generate undistinguishable pluripotent states and exclude a method-dependent epigenetic memory. Cell Culture. All the cell lines used in this study were derived from OG2 mice. Unless otherwise noted, all cell culture reagents were purchased from Gibco or PAA. Neonatal GSCs derived from 10 days postpartum (d.p.p) murine testis and adult GSCs derived from 35 d.p.p murine testis were derived as previously described and cultured on irradiated C3H feeder cells in a medium composed of StemPro-34 SFM (including supplement), N2 supplement (1:100), 1 % heat-inactivated fetal calf serum (FCS), 5 mg ml-1 bovine serum albumin (BSA) fraction V, 100 U ml-1 penicillin, 100 µg ml-1 streptomycin, 2 mM L‑glutamine, 0.1 mM non-essential amino acids (NEAA), 0.1 mM β‑mercaptoethanol, MEM vitamins (1:100), 6 mg ml-1 D‑(+)‑glucose (Sigma), 30 mg ml-1 pyruvic acid, 0.1 % DL‑lactic acid (Sigma), 60 ng ml-1 progesterone (Sigma), 30 ng ml-1 β‑estradiol (Sigma), 15 ng ml-1 mouse EGF (Peprotech), 10 ng ml-1 human bFGF (Peprotech), 10 ng ml-1 human GDNF (Peprotech), and 1000 U ml-1 mouse LIF (in‑house preparation). For maintenance culture, GSCs were passaged at a 1:2‑ratio once a week after digestion with 0.05 % trypsin/EDTA. In addition, ESCs, iPSC and gPSC were cultured on irradiated C3H feeders in the following medium: Knockout DMEM, 20% Knockout Serum Replacement, penicillin, streptomycin, L‑glutamine, NEAA, β‑mercaptoethanol, and LIF. Generation of pluripotent cell lines. iPSC were generated using tetracycline-inducible lentiviruses. For the lentiviral production, 293T cells were plated at 50 % confluency in 10-cm dishes one day prior to transfection. The next day, the lentiviral vector (3 µg) together with the packaging plasmids psPAX2 (2 µg) and pMD2.G (1 µg) (Addgene plasmids #12260 and #12259, respectively) were transfected using FuGENE 6 reagent (Roche) according to the manufacturer’s protocol. After 48 h, viral-containing supernatants were filtered (0.45 µm), concentrated around 30-fold by centrifugation (26,000 × g, 2 h, 4 °C) and stored at -80 °C. For the generation of iPSCs, a single-cell suspension of 50,000 GSCs was infected overnight with five tetracycline-inducible lentiviruses (5–20 µl concentrated supernatant each) encoding for Oct4, Sox2, Klf4, c-Myc, and the reverse tetracycline-transactivator M2-rtTA (Addgene plasmids #20323, #20326, #20322, #20324, and #20342) in the presence of 4 µg ml-1 polybrene (Sigma). One week after the infection, the expression of the viral transgenes was induced by adding 4 µg ml-1 doxycycline (Sigma) to the medium. MEFs-derived iPSCs were generated using the same protocol. gPSC were generated following a previous reported protocol. Briefly, GSCs were seeded at a density of approx. 1000 cells cm-2 on feeder cells and were not splitted until gPSC colonies were observed. Both iPSC and gPSC colonies were identified by Oct4-GFP expression, manually picked and expanded under ESC conditions. In vitro differentiation of pluripotent cells. Embryoid body (EB) formation was induced in suspension culture adopting the hanging drop method (600 cells in 20 µl) for 5 days in ESC medium without LIF. EBs were seeded on gelatine-coated plates and further cultured for 14 days in the same medium. Chimera formation. Aggregations of putative pluripotent cells with denuded 8-cell stage embryos were performed as previously reported. After overnight incubation at 37 °C and 5 % CO2, 11–14 aggregates were transferred into the uterine horns of 2.5 d.p.c pseudo-pregnant surrogate mothers (mated with vasectomized males). The chimeric pups were obtained and dissected at 14.5 d.p.c. Microarray gene expression analysis. Purified cRNA samples were prepared with the linear TotalPrep RNA Amplification Kit (Ambion) from 500 ng original RNA, involving synthesis of double-stranded cDNA by T7-promoter-linked oligo(dT) primers and 14 hours of in-vitro transcription incorporating biotin-labelled nucleotides. Hybridizations onto MouseRef 8 v2.0 Expression BeadChips (Illumina) was performed according to the manufacturer’s recommendations. Two replicates were hybridized from each sample. Chips were stained with streptavidin-Cy3 (GE Healthcare) and scanned using an iScan reader (Illumina). The bead intensities were mapped to gene information using BeadStudio 3.2 (Illumina). Background correction was performed using the Affymetrix robust multiarray analysis (RMA) background correction model. Variance stabilization was performed by log2 scaling, and gene expression normalization was calculated with the method implemented in the lumi package of R-Bioconductor.

Dataset Information

MethodsJ2: a software tool to capture metadata and generate comprehensive microscopy methods text.

Publications

MethodsJ2: a software tool to capture metadata and generate comprehensive microscopy methods text.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets