{"data":{"featured":{"edges":[{"node":{"frontmatter":{"title":"Scaling Laws in Single-Cell Representation Learning","cover":{"childImageSharp":{"gatsbyImageData":{"layout":"constrained","placeholder":{"fallback":"data:image/png;base64,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"},"images":{"fallback":{"src":"/sijiazhu.github.io/static/23a97d60552a128f13bb09db32741040/42837/cover.png","srcSet":"/sijiazhu.github.io/static/23a97d60552a128f13bb09db32741040/1ac8f/cover.png 175w,\n/sijiazhu.github.io/static/23a97d60552a128f13bb09db32741040/7d021/cover.png 350w,\n/sijiazhu.github.io/static/23a97d60552a128f13bb09db32741040/42837/cover.png 700w","sizes":"(min-width: 700px) 700px, 100vw"},"sources":[{"srcSet":"/sijiazhu.github.io/static/23a97d60552a128f13bb09db32741040/f0c3a/cover.avif 175w,\n/sijiazhu.github.io/static/23a97d60552a128f13bb09db32741040/39b02/cover.avif 350w,\n/sijiazhu.github.io/static/23a97d60552a128f13bb09db32741040/e6226/cover.avif 700w","type":"image/avif","sizes":"(min-width: 700px) 700px, 100vw"},{"srcSet":"/sijiazhu.github.io/static/23a97d60552a128f13bb09db32741040/f74b7/cover.webp 175w,\n/sijiazhu.github.io/static/23a97d60552a128f13bb09db32741040/aa3e5/cover.webp 350w,\n/sijiazhu.github.io/static/23a97d60552a128f13bb09db32741040/54028/cover.webp 700w","type":"image/webp","sizes":"(min-width: 700px) 700px, 100vw"}]},"width":700,"height":443}}},"tech":["Single-cell multi-omics","Random Matrix Theory","Representation learning"]},"html":"<p>This project studies how cell number, sequencing depth, and batch effects shape biological signal recovery in single-cell representation learning. The work uses mutual-information-based measures and real-data evaluation to connect theory with embedding performance.</p>"}},{"node":{"frontmatter":{"title":"Survival Analysis in Oncology Clinical Trials","cover":{"childImageSharp":{"gatsbyImageData":{"layout":"constrained","placeholder":{"fallback":"data:image/png;base64,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"},"images":{"fallback":{"src":"/sijiazhu.github.io/static/6f0455778ac57507ffdae340b04a8aed/42837/cover.png","srcSet":"/sijiazhu.github.io/static/6f0455778ac57507ffdae340b04a8aed/1ac8f/cover.png 175w,\n/sijiazhu.github.io/static/6f0455778ac57507ffdae340b04a8aed/7d021/cover.png 350w,\n/sijiazhu.github.io/static/6f0455778ac57507ffdae340b04a8aed/42837/cover.png 700w","sizes":"(min-width: 700px) 700px, 100vw"},"sources":[{"srcSet":"/sijiazhu.github.io/static/6f0455778ac57507ffdae340b04a8aed/f0c3a/cover.avif 175w,\n/sijiazhu.github.io/static/6f0455778ac57507ffdae340b04a8aed/39b02/cover.avif 350w,\n/sijiazhu.github.io/static/6f0455778ac57507ffdae340b04a8aed/e6226/cover.avif 700w","type":"image/avif","sizes":"(min-width: 700px) 700px, 100vw"},{"srcSet":"/sijiazhu.github.io/static/6f0455778ac57507ffdae340b04a8aed/f74b7/cover.webp 175w,\n/sijiazhu.github.io/static/6f0455778ac57507ffdae340b04a8aed/aa3e5/cover.webp 350w,\n/sijiazhu.github.io/static/6f0455778ac57507ffdae340b04a8aed/54028/cover.webp 700w","type":"image/webp","sizes":"(min-width: 700px) 700px, 100vw"}]},"width":700,"height":443}}},"tech":["Survival analysis","Cox modeling","Clinical trials"]},"html":"<p>An applied biostatistics project focused on multi-arm oncology clinical trial data. The analysis constructs time-to-event endpoints, handles staggered treatment initiation and censoring, and evaluates treatment effects with Kaplan-Meier and Cox proportional hazards models.</p>"}},{"node":{"frontmatter":{"title":"Pathway-Independent Genetic Signals in T2D Clusters","cover":{"childImageSharp":{"gatsbyImageData":{"layout":"constrained","placeholder":{"fallback":"data:image/png;base64,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"},"images":{"fallback":{"src":"/sijiazhu.github.io/static/0a28dbd7331562f51f6eb76aab8cb0b2/42837/cover.png","srcSet":"/sijiazhu.github.io/static/0a28dbd7331562f51f6eb76aab8cb0b2/1ac8f/cover.png 175w,\n/sijiazhu.github.io/static/0a28dbd7331562f51f6eb76aab8cb0b2/7d021/cover.png 350w,\n/sijiazhu.github.io/static/0a28dbd7331562f51f6eb76aab8cb0b2/42837/cover.png 700w","sizes":"(min-width: 700px) 700px, 100vw"},"sources":[{"srcSet":"/sijiazhu.github.io/static/0a28dbd7331562f51f6eb76aab8cb0b2/f0c3a/cover.avif 175w,\n/sijiazhu.github.io/static/0a28dbd7331562f51f6eb76aab8cb0b2/39b02/cover.avif 350w,\n/sijiazhu.github.io/static/0a28dbd7331562f51f6eb76aab8cb0b2/e6226/cover.avif 700w","type":"image/avif","sizes":"(min-width: 700px) 700px, 100vw"},{"srcSet":"/sijiazhu.github.io/static/0a28dbd7331562f51f6eb76aab8cb0b2/f74b7/cover.webp 175w,\n/sijiazhu.github.io/static/0a28dbd7331562f51f6eb76aab8cb0b2/aa3e5/cover.webp 350w,\n/sijiazhu.github.io/static/0a28dbd7331562f51f6eb76aab8cb0b2/54028/cover.webp 700w","type":"image/webp","sizes":"(min-width: 700px) 700px, 100vw"}]},"width":700,"height":443}}},"tech":["Type 2 diabetes","Polygenic scores","Genetic clustering"]},"html":"<p>This project investigates whether type 2 diabetes genetic clusters contain residual biological signals beyond curated pathway-based polygenic scores from resources such as Reactome and KEGG.</p>"}}]}}}