# Search log — exploratory sweep behind the first 138 papers Run Wed 7 Oct 2026 (NZDT). Two parallel searches over open bibliographic APIs. Not the PRISMA-ScR search; see the Method page for what the formal search will be. ## Search A — behavioural science, implementation science, population & cohort, cross-domain analogues **(d) Searches and databases (all run 7 Oct 2026)** - **Europe PMC REST API** (`/webservices/rest/search`, resultType=core): exact-title and field queries for every SoRI-cited paper (e.g. `TITLE:"digital health interventions" AND "physical activity" AND "community-dwelling" AND AUTH:"Di Pumpo"`; `MHC-Coach transtheoretical model large language model`); `TITLE:"novelty effect" AND (wearable OR "activity tracker" OR mHealth)`; `TITLE:"wearable activity trackers" AND (TITLE:"systematic review" OR TITLE:"meta-analysis") AND "older adults"`; `TITLE:"gamification" AND TITLE:"physical activity" AND TITLE:"meta-analysis"`; `TITLE:"gamification" AND TITLE:"older adults" AND TITLE:"physical activity" AND randomized`; `TITLE:"reinforcement learning" AND TITLE:"physical activity"`; `TITLE:"large language model" AND TITLE:"physical activity" AND (coach OR coaching)`; `TITLE:"chatbot" AND TITLE:"physical activity" AND (systematic review OR meta-analysis)`; `TITLE:"financial incentives" AND TITLE:"physical activity" AND TITLE:"meta-analysis"`; `TITLE:"Māori" AND (digital health OR mHealth OR telehealth)`; `TITLE:"dark patterns" AND (health OR app)`; `TITLE:"gamification" AND (ethics OR ethical) AND health`; `TITLE:"older adults" AND TITLE:"wearable" AND (abandon OR long-term use OR sustained use)`; `TITLE:"habit" AND TITLE:"physical activity" AND (meta-analysis OR systematic review)`; `TITLE:"exergames" AND TITLE:"older"`; `TITLE:"adaptive" AND TITLE:"goal" AND TITLE:"physical activity"`; `TITLE:"HeartSteps" AND micro-randomized`; plus exact-title checks for Teixeira 2012, Ntoumanis 2021, Hermsen 2017, Patel 2015/2019/2021, Jakicic 2016, Meyerowitz-Katz 2020, Peters 2018, Hardeman 2019, Greenhalgh 2017, Michie 2013, Bowen 2009, Gardner 2012, Bickmore 2013, Finkelstein 2016. Full-text XML for Di Pumpo (PMC11738617) checked for the 7,068-record figure. - **OpenAlex** (`/works?search=`): PhysioLLM; Engage-PA; self-concordance older adults; LumiHealth; Vitality Active Rewards Apple Watch; Healthier SG; SDT (Ryan & Deci 2000); BCW; MINDSPACE; Lally 2009; Nahum-Shani JITAI; Yardley 2016; Perski 2016; Eysenbach 2005; RE-AIM; CFIR; Csikszentmihalyi flow. (Rate-limited partway; remaining queries moved to Europe PMC/Crossref.) - **Crossref** (`/works?query.bibliographic=`): Fogg 2009; Sheldon & Elliot 1999; Hunicke 2005; Hamari 2014; Deterding 2011; Sunstein 2020; Ryan, Rigby & Przybylski 2006; Prochaska & Velicer 1997; Nakamura & Csikszentmihalyi; Mathur 2019; Schüll 2012; Koivisto & Hamari 2019; Gray 2018; Jörke 2025 GPTCoach; Di Pumpo publication date (`/works/10.3389/ijph.2024.1607720` → published 3 Jan 2025, vol 69). - **Web search / fetch**: "Swiss Re LumiHealth analysis physical activity Apple Watch Singapore" → HPB newsroom (fetched) and Swiss Re PDF (fetched, text extracted); "MINDSPACE Institute for Government 2010 … EAST Behavioural Insights Team 2014" → IfG and BIT PDF URLs (not fetched; [METADATA]); "Health New Zealand Te Whatu Ora digital health strategy Māori equity Hira" → secondary news only (excluded). ## Search B — LLM/AI, ML state estimation, feedback & control, wearables, ethics & embodiment ## (d) Search strings and databases (all run Wed 7 Oct 2026, NZDT) - **OpenAlex** (`api.openalex.org/works?search=`): about 75 queries, e.g. "Just-in-time adaptive interventions in mobile health key components and design principles"; "HeartSteps micro-randomized trial physical activity suggestions Klasnja"; "control systems engineering behavioral interventions Rivera"; "adaptive step goals randomized trial physical activity Adams"; "large language model fine-tuned transtheoretical model physical activity coaching My Heart Counts"; "PhysioLLM wearable data large language model personalized health insights"; "Apple Intelligence foundation language models tech report 2025"; "ElliQ older adults social robot evaluation"; "LOVOT robot older adults loneliness"; "large language models personal health insights wearable data Gemini PH-LLM"; "receptivity just-in-time interventions mobile health prediction"; "conversational agents physical activity diet behaviour change systematic review meta-analysis"; "LLM generated motivational messages physical activity evaluation"; "Apple Watch heart rate variability validation"; "Apple Watch VO2max estimation validity"; "wearable activity trackers older adults use and abandonment"; "weather and physical activity accelerometer older adults"; "novelty effect wearable fitness tracker engagement decline"; "dynamic difficulty adjustment exergame older adults"; "digital nudging autonomy manipulation ethics"; "socially assistive robots older adults loneliness systematic review meta-analysis"; "Just Walk system identification physical activity intervention"; "large language model conversational agent older adults"; "QLoRA efficient finetuning quantized LLMs". From mid-session the API returned HTTP 429 (rate limit), so later searches moved to Crossref and Europe PMC. - **Crossref** (`api.crossref.org/works?query.bibliographic=` and `/works/{DOI}`): DOI checks for Klasnja 2019, Fogg 2009, Liao 2020, Korinek 2018, Rieke 2020, Yardley 2016 and Perski 2017. Bibliographic queries for Yerrakalva 2019, Mishra 2021, Robinson 2013, Broadbent 2014, Meskó & Topol 2023, Engelen 2019, Brickwood 2019, Attig & Franke 2020, Clawson 2015, Lazar 2015, Riley 2011, Hekler 2018, Zhou 2018, Hunicke 2005, Aguilera 2024, Bickmore 2005/2013, Phatak 2018, King 2007/COMPASS, CoachAI, Looi, Okeke and HeartSteps outcomes. - **Europe PMC** (`ebi.ac.uk/europepmc/webservices/rest/search`, resultType=core): abstract retrieval by DOI for about 25 trials and reviews. Free-text queries: "conversational agent older adults lifestyle", "contextual bandit physical activity messages personalization trial", "sedentary interruption JITAI older adults wearable", "ElliQ older adults living alone loneliness", "Maori perspectives on privacy". - **arXiv API** (`export.arxiv.org/api/query?id_list=`): 1802.07740 (Rabinowitz), 1909.03539 (Liao), 2506.07275 (Song), 2312.06674 (Llama Guard). - **Publisher and official pages:** proceedings.mlr.press/v80/rabinowitz18a.html (title confirmed); privacy.org.nz HIPC 2020 page (title confirmed). - Semantic Scholar was attempted and returned HTTP 429. It was not used. ## Afterwards - Rows deduplicated by DOI: 76 (A) + 71 (B) → 135 unique. - Added 8 Oct 2026 after Crossref verification: Intille 2004; Spruijt-Metz & Nilsen 2014; Nahum-Shani, Hekler & Spruijt-Metz 2015 (JITAI lineage) → 138. - GPTCoach (Jörke et al. 2025) recoded 8 Oct 2026 from the arXiv full text (D3, D4, D5). - Inclusion rule for the sweep: a DOI, arXiv id or stable URL was resolved and the title seen; provenance per row: FULL-TEXT-READ, ABSTRACT-READ or METADATA. ## Pilot of the planned search — open databases, 13 Oct 2026 Run 13 Oct 2026 08:02 NZDT. Three concept blocks AND-ed (population 50+/older · digital intervention · physical activity), plus a 4-block variant adding the outcome block (engagement, adherence, retention, attrition, "novelty effect", long-term, sustained, maintenance, abandonment, "continued use", dropout). Limits 2010–2026; English (Europe PMC). Exact strings and the per-record log are in the candidate workbook (`_search-candidates-2026-10-13.xlsx`, sheets Summary and Log) and in `_run_search.py`. | Database | Query | Hits | Pulled | |---|---|---|---| | Europe PMC (REST, TITLE/ABSTRACT fields) | 3 blocks | 2,728 | 2,728 | | Europe PMC | 4 blocks (+ outcome) | 833 | 833 | | OpenAlex (`title_and_abstract.search`) | 3 blocks | 18,338 | 3,000 most cited | | OpenAlex | 4 blocks (+ outcome) | 6,083 | 6,083 | | arXiv API | (LLM OR chatbot OR conversational agent) AND (PA OR exercise OR walking OR fitness) AND (coach* OR intervention OR behaviour change OR older adults) | 119 | 119 | | arXiv API | (just-in-time adaptive OR JITAI OR micro-randomized OR contextual bandit) AND (PA OR mobile health OR mHealth OR steps) | 146 | 146 | | ClinicalTrials.gov v2 | PA × digital × older terms; MinimumAge ≥ 50 y; StartDate ≥ 2010 | ≥ 1,000 | 1,000 (cap) | Flow: 12,909 identified → 9,657 unique (DOI, then normalised title) → 27 already charted → 9,630 new. Machine pre-screen (regex on title + abstract; a sorting aid, not a decision): tier A 2,674 (strict 1,455), tier B 208, reviews 595. Read at abstract level: the 55 strict-A records in the 50+ × (D5|D6|D7) cell (33 charted) and the 26 dimension-carrying records among 54 provisional "in" from a pre-screened 275-record 50+ × ≥6-month batch (23 charted). Table 135 → 191 rows. Sensitivity: Bickmore 2013, Greysen 2024, King 2016, Patel 2019, Finkelstein 2016, Hermsen 2017 retrieved; Wong 2024 NOT retrieved (abstract uses "wearable monitoring device" / "healthy behaviors", no PA term) → behaviour block to gain tracker-use / health-behaviour terms before the formal run. Not run here: MEDLINE (Ovid), Embase, PsycInfo, CINAHL, Scopus, Web of Science, IEEE Xplore, ACM DL, ANZCTR — subscribed/manual, campus PC.