Anu Ramachandran

Anu Ramachandran

Auto-generated profile, based on publicly available information

Anu Ramachandran

Center for Innovation to Implementation, VA Palo Alto, Palo Alto, CA, USA.

Research Focus & Metrics

Academic Career Length

2Years (2025-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
15 March 2026 in Conflict and Health
Verified
Journal Article
Open access
ARAnu Ramachandran
Anu Ramachandran
STSu Mon Thwe
Su Mon Thwe
CWCho Zin Win
Cho Zin Win
NHNay Lynn Htet
Nay Lynn Htet
SMSaw Kyaw Myint
Saw Kyaw Myint
PPParveen Parmar
Parveen Parmar
Background: Cardiovascular disease (CVD) is a leading cause of death in low-income countries and those affected by armed conflict, including Myanmar. Community health worker interventions can effectively address CVD risk factors in low-income countries but have not been tested among displaced populations in active conflict zones. Objectives: We conducted a feasibility study of a village health worker (VHW) care model to identify individuals at high CVD risk and deliver care in conflict-affected regions of Karen State, Myanmar. This study was conducted by an international non-governmental organization collaborating with a regional local health organization. Methods: Following a village census, trained VHWs and medics screened individuals age ≥ 40 for CVD risk factors in three villages. Eligible individuals had HTN, diabetes, calculated CVD risk > 10%, or history of heart attack or stroke, confirmed during a second visit 1–2 weeks later. VHWs visited households every 3–6 weeks for 2 months to monitor blood pressure, glucose, medication adherence/side effects, and deliver medic-prescribed medications. Feasibility evaluation centered on reach, adoption, and acceptability. Outcomes included CVD risk factor prevalence, recruitment and retention, medication initiation/adherence, changes in hypertension control, and adverse outcomes. VHW and medic focus group discussions and study participant interviews were conducted. Results: CVD teams screened 294 individuals, conducted confirmatory visits with 132, enrolled all 97 eligible participants, and completed two home visits with 94 patients. Several prescription errors were made, halting medication initiation; root cause analysis identified opportunities to improve pre-testing of electronic tools and strengthen clinician CVD training. The proportion of eligible participants receiving antihypertensive or statin medications increased from 23% to 56%. Among those with HTN, the proportion achieving blood pressure control < 140/90 mmHg increased from 22.9% to 65.7%. Qualitative assessment revealed support for the care model and opportunities to improve training and streamline clinical protocols. Conclusions: Our study suggests a VHW care model for CVD in remote villages in Myanmar experiencing armed conflict is feasible and can increase medication access. Opportunities exist to simplify CVD treatment guidelines and augment training and support of local providers. Findings informed a cluster randomized controlled trial to test the impact of a modified VHW care model on medication adherence, CVD risk, and cost. Trial registration: ClinicalTrials.gov, NCT06819839, retrospectively registered 27 October 2024.

27 October 2025 in PLOS Digital Health
Verified
Journal Article
Open access
PDF file
ARAnu Ramachandran
Anu Ramachandran
AYAkash Yadav
Akash Yadav
ASAndrew Irwin Schroeder
Andrew Irwin Schroeder
Earthquakes and other disasters often cause substantial damage to health facilities, impacting short-term response capacity and long-term health system needs. Identifying health facility damage following disasters is therefore crucial for coordinating response, but ground-based evaluations require substantial time and labor. Artificial intelligence (AI) models trained on satellite imagery can estimate building damage and could be used to generate rapid health facility damage reports. There is little published about methods of generating these estimates, testing real-world accuracy, or exploring error. This study presents a novel method of overlaying model damage outputs with health facility location data to generate health facility damage estimates following the February 2023 earthquake in Turkey. Two models were compared for agreement, accuracy, and errors. Building-level damage estimates were obtained for Model A (Microsoft neural network model), and Model B (Google AI model), and overlaid with health facility location data to identify facilities with significant damage. Model agreement, sensitivity and specificity for damage detection were calculated. A descriptive review of common error sources based on selected satellite imagery was conducted. A spatially aggregated damage estimation, based on proportion of buildings damaged in a 0.125km 2 area, was also generated and assessed for each model. Twenty-five hospitals, 13 dialysis facilities, and 454 pharmacies were evaluated across three cities. Estimated damage was higher for Model A (10.4%) than Model B (4.3%). Cohen’s kappa was 0.32, indicating fair agreement. Sensitivity was low for both models at 42.9%, while specificity was high (A:93.6%, B:96.8%). Agreement and sensitivity were best for hospitals. Common errors included building identification and underestimation of damage for destroyed buildings. Spatially aggregated damage estimates yielded higher sensitivity (A:71.4%, B:57.1%) and agreement (Cohen’s kappa 0.38). Leveraging remote-sensing models for health facility damage assessment is feasible but currently lacks the sensitivity to replace ground evaluations. Improving building identification, damage detection for destroyed buildings, and spatially aggregating results may improve the performance and utility of these models for use in disaster response settings.

13 November 2026 in medRxiv
Verified
Preprint
Open access
PDF file
ZSZili Shen
Zili Shen
ARAnu Ramachandran
Anu Ramachandran
EHElaine Hsiang
Elaine Hsiang
JGJasleen Gill
Jasleen Gill
RMRyan Miller
Ryan Miller
NSNicholas Stark
Nicholas Stark
Agentic healthcare platforms that combine large language models with physician oversight are growing quickly, but little is known about who uses them. We characterized 3,223 enrollees of Lotus AI, a free, physician-supervised agentic platform, who completed onboarding, messaged the agent, and had a linked health record between February and July 2026, and compared them with national benchmarks. Median age was 51 years and 62% were male. Chronic condition prevalence exceeded 2024 National Health Interview Survey estimates in every age band by 4 to 7 percentage points; 18% had a functional or symptom-defined disorder; among the 30% with four or more ambulatory visits in the prior year, care was spread across nearly five providers, with 44% meeting a threshold for high fragmentation. Users with functional disorders generated 1.8 times as many physician escalations and 2.6 times as many laboratory orders as other users, but no more prescriptions. This profile differs from the younger, predominantly female populations reported for direct-to-consumer telemedicine. It moves chronic disease management, functional disorder pathways, and care coordination from aspirations to design requirements, and we argue that deployed AI care platforms should characterize their populations against national benchmarks early.