The rapid evolution of the Internet of Things has led to the development of smart health. As a form of medical care that uses advanced Internet technology to realize better diagnosis and treatment of patients, smart health transitions medical services move toward real intelligence and greatly helps users. And in smart health, the secure sharing of personal health records (PHRs) is one of the main concerns of patients and medical personnel. Many attribute-based sharing models have been proposed to secure the sharing of PHRs, but there are still two problems to resolve. One is the potential disclosure of the patient data. The attribute-based model achieves flexible access control, but the access policies contain sensitive information of patients. The disclosure of the policy will lead to the leakage of data of the users. The other is the high computational and storage overhead, particularly in smart health systems with limited computing power. In this article, we present a Smart Health-Lightweight Fine-Grained Sharing (SH-LFGS) framework based on attribute-based encryption (ABE). It achieves a fully hidden access policy by adopting Viéte's formula. SH-LFGS introduces an online/offline mechanism in the PHR encryption phase and the outsourced verifiable decryption mechanism. Because the decrypting test requires only one bilinear pair operation, the SH-LFGS can achieve the task of lightweight computation. Analysis of the performance and security of the proposed model confirm its efficiency and security.