内容来源：https://www.livemint.com/ai/healthtech-startups-ai-electronic-health-records-urban-hospitals-rural-clinics-qure-ai-jivi-ai-tulu-health-healthplix-11746890149900.html

内容总结：

印度医疗科技初创企业正积极运用人工智能技术构建电子健康记录（EHR）系统，旨在打通城乡医疗数据壁垒，但实际应用推广仍面临多重挑战。

目前，Qure.ai、医准智能（Jivi.ai）、Tulu Health、HealthPlix、Eka.care等企业正开发AI驱动工具，通过语音转录、智能问诊提示等功能将医患交互内容转化为结构化病历。不同于仅限单家机构使用的电子病历（EMR），EHR系统可实现跨机构数据互通，对提升诊疗效率和连续性具有关键意义。

尽管技术前景广阔，印度EHR系统的普及率仍处于较低水平。数据显示，全国14亿人口中仅57%注册了阿育曼数字健康计划（ABDM）账户，实际关联健康记录的比例更低。城乡数字化基建不均衡、医疗机构改造成本高昂、缺乏专业IT团队以及数据隐私法规待完善等因素，共同制约了技术落地。

值得注意的是，农村地区展现出强烈需求但实施难度更大。Eka.care等企业通过开发离线模块、实时语音转文本工具应对网络瓶颈，但大规模部署仍存障碍。投资者对该领域保持审慎态度，强调技术方案需兼具合规性、可扩展性及明确投资回报路径。

行业专家指出，医疗机构对引入AI系统存在成本与责任顾虑，且升级传统系统资金投入巨大。目前多数企业处于概念验证阶段，能否突破医院端规模化应用仍有待观察。随着印度《数字个人数据保护法（2023）》生效，数据安全规范将为行业发展提供新框架。

中文翻译：

印度医疗科技初创企业正竞相利用人工智能技术实现医疗记录数字化，服务范围涵盖城市医院与乡村诊所。这些企业瞄准了医疗系统中的关键短板——缺乏强健且可互操作的电子健康档案（EHR）。以人工智能为核心，Qure.ai、Jivi.ai、Tulu Health、HealthPlix、Eka.care、Primera Medical Technologies和KareXpert等公司正在开发工具，将医患互动数字化并生成结构化医疗记录。

与仅限于单个医院或诊所的电子医疗记录（EMR）不同，电子健康档案的设计可实现跨机构调阅。这种互操作性对提升诊断、治疗和护理连续性至关重要。更重要的是，私立和公立医疗机构都能访问这些记录。

Qure.ai的语音助手Aira能转录并总结患者对话，提示医护人员提出更精准的问题以改善诊断。"我们发现解决电子医疗记录的途径之一是将医患对话数字化，"公司创始人兼首席执行官普拉尚特·瓦里尔表示。Aira目前正在尼日利亚、瓦拉纳西和马哈拉施特拉邦进行试点，计划推广至印度公共卫生中心。

该公司的AI工具旨在支持中低收入国家的一线医疗工作者。"人工智能能有效扩充他们的知识库，这将成为颠覆性的解决方案，"首席产品官安基特·莫迪指出。Qure的客户包括阿斯利康、英国国家医疗服务体系及东南亚多国卫生部。在Lightspeed、Peak XV和默克全球健康创新基金的支持下，公司2024财年收入增至14亿卢比，亏损收窄至4.8亿卢比。

在农村地区，Eka.care通过与医生、患者和医院合作构建电子医疗记录。公司创始人维卡尔普·萨尼表示："我们看到乡村医疗机构、社区卫生所和企业社会责任项目对数字化医疗的需求日益增长。"为应对网络带宽不足的挑战，公司开发了AI语音转文本工具Eka Scribe，能实时将医患对话转化为结构化处方，并开发了离线优先模块，可在网络恢复后自动同步数据。

同样获得前谷歌大脑负责人吴恩达AI基金支持的Jivi.ai，正尝试进入印度和美国医院工作流程，通过应用采集病史生成医疗记录。创始人安库尔·贾恩透露，公司目前与多家医院处于概念验证阶段，但因销售周期长暂未公开合作方信息。

投资者对电子医疗记录领域态度谨慎。Quadria资本合伙人苏尼尔·塔库尔指出："我们寻求符合规范、可扩展的SaaS技术供应商，要求其具备清晰的投资回报路径和行为适应机制。"专家认为大规模部署仍存挑战，普华永道全球医疗行业顾问苏杰·谢蒂表示："医院对引入AI存在成本顾虑，且升级传统系统需巨额投入。"

监管方面，印度尚未出台医疗数据专项法案，相关平台受中央药物标准控制组织监管，数据保护则遵循新出台的《数字个人数据保护法(2023)》。企业需通过沙盒环境测试、技术整合和安全审计后方可接入国家数字医疗使命计划。

Qure.ai的瓦里尔透露，语音转录系统在高噪音环境下表现良好，但精准提示功能仍需优化："提供最准确提示仍需改进，这是我们当前的工作重点。"

英文来源：

Healthtech startups bet on AI to build electronic health records but adoption in India remains slow
Indian healthtech startups are racing to digitize medical records using artificial intelligence, targeting both urban hospitals and rural clinics.
India’s healthtech startups are zeroing in on a critical gap in the healthcare system: the lack of robust, interoperable electronic health records (EHRs). With artificial intelligence as their backbone, companies like Qure.ai, Jivi.ai, Tulu Health, HealthPlix, Eka.care, Primera Medical Technologies and KareXpert are building tools to digitise doctor-patient interactions and create structured medical records.
Unlike electronic medical records (EMRs), which are confined to individual hospitals or clinics, EHRs are designed to be accessible across institutions. This interoperability is key to improving diagnosis, treatment, and continuity of care. What's more, they can be accessed across both privately held and government-operated healthcare providers.
Qure.ai’s voice-based assistant, Aira, transcribes and summarizes patient conversations, nudging healthcare workers to ask better questions and improve diagnosis. “We realized one way to solve for EMRs is by digitising the doctor-patient conversation," said Prashant Warrier, founder and CEO of Qure.ai. Aira is currently being piloted in Nigeria, Varanasi, and Maharashtra, with plans to expand to Indian public health centres.
Based on the conversation with the patient, Qure's AI ‘nudges’ healthcare workers, improving the quality of their questions while simultaneously helping provide better diagnoses and remedies.
The company’s AI tools is meant to support frontline workers like Ashas in LMICs. “Here, AI can basically augment their knowledge base; it becomes a game-changing solution," said Ankit Modi, Qure’s chief product officer.
Qure’s clients include AstraZeneca, the UK’s National Health Service, and ministries of health in Southeast Asia. Backed by Lightspeed, Peak XV, and Merck’s Global Health Innovation Fund, Qure’s FY24 revenue rose to ₹140 crore, narrowing losses to ₹48 crore.
Rural areas
Similarly, Eka.care is a public health records app which works with doctors, patients and hospitals to build electronic medical records (EMRs). The company has been seeing a lot of interest in its products, especially in rural areas. “We’re seeing increasing interest from rural healthcare providers, community health clinics, and CSR-backed initiatives who want to digitize care at the last mile," said company founder and CEO Vikalp Sahni in a written response toMint.
But rural India brings its own set of challenges, like low bandwidth and poor connectivity. To combat that, Eka.care built a similar tool to Qure's Aira, which it calls Eka Scribe, an AI-powered voice-to-text tool that converts conversations between doctors and patients into structured prescriptions in real time. The company is also working on offline-first modules, which will then auto-sync to create records once bandwidth access is restored. “The goal is to meet doctors where they are, whether it's OPDs or if they're on the move without compromising on medical accuracy or record quality," Sahni said.
Eka.care last raised $15 million in a Series A round in 2022, led by Hummingbird Ventures. The company's earliest investors included deeptech-focused Speciale Invest and early-stage VC firm 3one4 Capital. As of September 2022, its valuation was $49.8 million, according to Tracxn.
Similarly, Jivi.ai, backed by former Google Brain head Andrew Ng's AI fund, is trying to make diagnostics possible and take it a step further.
The company is trying to enter hospital workflows, both in India and the US, where users can have their history taken by the app before connecting them with doctors, thereby creating medical records. “Our strategy is to focus on very large-scale healthcare systems and work with them to solve for a particular use case," said Jivi founder and chief executive Ankur Jain.
Currently, most of its attempts are in the proof-of-concept (PoC) stage with various hospitals across India and the US. However, Jain declined to share names of companies they're working with, saying that they were still at the PoC stage and that sales cycles were long.
Investor interest
In India, investor sentiment for companies working on building EMRs and EHRs is cautious, given the low adoption of the technology in hospitals, even in urban ones. Then there's the rural angle, where hospitals outside metros lack high-speed internet, trained IT staff, and information and communication infrastructure. This, in turn, makes it harder to deploy EMR solutions.
“We seek EMR startups that offer compliant, scalable, and SaaS-based technology. They should provide clear ROI pathways, behavioural adoption mechanisms, and additional services. Ideally, these startups should cater to large institutional clients," said Sunil Thakur, partner and investment committee member at Quadria Capital, a healthcare-focused private equity firm which invests across South and Southeast Asia.
State of electronic health records in India
The government began working on creating EHRs in 2017 under the National Health Policy. Under it and the National Digital Health Blueprint, the government launched the National Digital Health Mission (now known as Ayushman Bharat Digital Mission, or ABDM).
Since then, adoption has dramatically picked up, but it's still lagging.
According to the ABDM dashboard created by the government and last updated on 1 September, the total number of Ayushman Bharat Health Accounts (Abha) stands at 804 million. India's population is 1.4 billion, which means that only about 57% of the population has signed on to participate in the ABDM ecosystem.
Meanwhile, the number of health records linked with Abhas is even lower, at 714 million. Government hospitals account for 55% of those registered with the ABDM, while private hospitals make up 45%.
The path forward
Large-scale deployment of software for the creation of EHRs or even EMRs, for that matter, continues to remain an issue, according to experts. “I'm not seeing anyone at the enterprise level introducing such processes. However, at the startup level, we see people trying to do PoCs to show what they can do," said Sujay Shetty, global health industries advisory leader at PwC.
Some of this can be attributed to hospitals being wary about AI in their ecosystems because its introduction brings costs and liabilities. What's more, creating a data infrastructure that can ‘converse’ with AI is expensive. “It's also a function of the appetite of an institution on how much they want to spend to modernise legacy systems," said Shetty.
Regulations
Notably, India doesn't have a healthcare-specific law regarding patient data and privacy, unlike the US, which has its own version, the Health Insurance Portability and Accountability Act (Hippa). Platforms like Qure.ai and Eka.care are regulated by India's Central Drugs Standard Control Organisation. Data protection, on the other hand, falls under the incoming Digital Personal Data Protection Act (2023).
Qure takes the doctor's consent to record and transcribe conversations. Eka.care's Sahni said that their AI products are built with a "consent-first" design in mind. "From Abha-linked records to DocAssist AI suggestions, every action is logged, transparent, and controlled by the doctor or the patient."
Qure, for example, had to receive Food and Drug Administration approval to sell in US, while in the European Union, the company had to self-certify with CE marking, a process where companies declare their products comply with EU standards.
Startups can access the ABDM programme by accessing a sandbox environment, integrating their technology with the applications within the ABDM ecosystem, and being audited by a committee for both functionality and security. Once companies pass this, they showcase a demo of their product before finally being allowed to go live and link their product with the ABDM ecosystem.
Further, Qure's Warrier said that the transcription programme was performing well, especially in high background noise or poor device quality conditions, but that there was scope for improvement for the 'nudge' aspect of Aira. “Providing the most accurate nudges still needs some work. It's good, but it can be more accurate, and that's what we're working on now."