Home Hongjun Intelligence Files for IPO: AI-Powered 60-Second Video Mental Health Assessment with 78% Accuracy

Hongjun Intelligence Files for IPO: AI-Powered 60-Second Video Mental Health Assessment with 78% Accuracy

Oct 07, 2023 08:00 CST Updated 08:00

Depression, anxiety disorders, bipolar disorder... As these terms frequently appear in the public eye, it signifies that mental health issues are becoming increasingly prevalent.

 

According to the World Health Organization, there are approximately 350 million people worldwide suffering from depression, with up to one million suicides attributed to the condition each year. The situation in China is also concerning. Data from the “2022 Blue Book on National Depression” indicates that more than 95 million people in China currently suffer from depression, meaning roughly one in every 14 individuals is affected. Mental health has now become a major global focus, and many countries, including China, the United States, the United Kingdom, Australia, and Canada, have enacted corresponding laws, regulations, or policies to safeguard the mental well-being of their populations.

 

However, two major contradictions in the mental health market have led to its severity being largely overlooked, with approximately 75% of patients with mental health conditions not receiving systematic diagnosis and treatment.First, there is a shortage of medical resources on the supply side, with a severe mismatch between huge market demand and resource availability; second, there are constraints on the demand side, where many patients hesitate due to "stigma" and cost concerns, often suffering from depression in silence.

 

Faced with enormous market demand and a significant supply gap, traditional psychological counseling and therapeutic approaches appear inadequate. Artificial intelligence (AI), as an emerging technological tool, holds promise for ushering in a new era in mental health services.

 

Against this backdrop, Hunan Hongjun Intelligent Technology Co., Ltd. (hereinafter referred to as “Hongjun Intelligent”) was established, primarily targeting the AI mental health market. The company focuses on mental health identification and assessment, developing non-contact psychological state analysis technology that leverages observable behavioral cues to assess implicit psychological traits.

 

 

60-Second Video Assessment of Mental Health Status: Fine-Grained Scanning Across Five Psychological Dimensions


Traditional psychological health assessments rely on scale-based evaluations, which assess an individual’s psychological traits, behaviors, and emotional states by collecting data from their responses to specific questions. Common scales include the Self-Rating Depression Scale (SDS) and the Self-Rating Anxiety Scale (SAS), with each scale representing a distinct assessment dimension. However, this approach is inherently subjective, as results depend entirely on the respondent’s self-reported answers. Many respondents struggle with ambiguous options such as “agree” versus “strongly agree” or “disagree” versus “slightly disagree,” often finding it difficult to make a definitive choice despite careful deliberation.

 

PsychoLogic, also known as Xiao Sai AI, derives its name from the English word “Psychology.” It is an AI-powered mental health assessment system developed by Hongjun Intelligence. Unlike traditional scale-based assessments, Xiao Sai AI allows users to obtain a mental health status report and improvement recommendations within minutes by simply recording a 60-second video. It simultaneously evaluates five psychological dimensions, offering a process as simple and effortless as measuring blood pressure.

 

Xiao Sai AI's intelligent psychological state analysis technology is based on the "cognition, emotion, and somatic" theory of psychological state assessment. It integrates “AI + Image Recognition + Psychiatry and Psychology”, constructing intelligent analysis algorithms and recognition models. Meanwhile, by employing machine learning algorithms such as multi-task learning and incremental regression, we integrate various behavioral features with the Naive Bayes algorithm for decision fusion, thereby enhancing the accuracy of psychological indicators and emotion recognition models. To improve model robustness and generalizability, cross-validation is also used to evaluate model accuracy.

 

The dimensions of its assessment includeAnxiety, Stress, Self-Esteem, Achievement Drive, AI Mental Health Index

 

Anxiety: A state of irritability resulting from excessive worry about life safety, future prospects, and destiny.

Stress: The process of adaptive and coping responses elicited when environmental changes are perceived as threatening or challenging to the organism;

Self-Esteem: The outcome of an individual's self-evaluation of their social role;

Achievement-Driven: A desire to engage in adventurous activities or adopt unconventional lifestyles to seek thrilling experiences;

AI Mental Health Index: The degree of low mood, slowed thinking, and reduced and retarded speech and movement.

 

The AI Mental Health Index is scored in four tiers: below 37, 37–65, 65–75, and above 75. The other four dimensions are each scored in three tiers, with higher scores indicating greater severity.

 

 

AI-Extracted Key Points and Psychological Dimension Fitting Achieve 78% Accuracy in the Launched Version


Users access the assessment via a WeChat Official Account and then easily complete the required task by recording a 60-second video. Upon successful upload, the system automatically analyzes the user’s behavioral data, including audio metrics such as rhythm, spectrogram, and sound quality, as well as video-based metrics involving facial expressions, finger movements, and torso posture. During data recognition, the backend processes an amount of data equivalent to approximately 3,000 images. This process can be specifically divided into four steps:

 

I. Extract Key Points

Analyze facial features during reading to obtain information such as the curvature of the eye corners and mouth corners. During this process, AI extracts users’ facial keypoints and behavioral data through neural networks to achieve dimensionality reduction of the information.

 

II. Construction of an AI Psychological Trait Model

By analyzing facial representations of different individuals to construct psychological trait models, and simultaneously building a modeling vector database based on evidence-based data, we ensure that the model construction is reproducible, evidence-based, and trustworthy.

 

III. Effective Prediction of Psychological States

Based on accurate keypoint extraction and AI feature models, the assessment results are precisely fitted to psychological dimensions, enabling effective prediction and evaluation of users' psychological states.

 

IV. Generate Evaluation Report

Predict users' mental health status based on assessment results, and generate a psychological dimension scoring report along with improvement recommendations.

 

Overall, Xiao Sai AI offers advantages such as convenience, fast speed, high efficiency, greater objectivity, and strong privacy protection.

 

On May 30, 2023, Hongjun Intelligent’s AI-based psychological assessment software obtained a Class II Medical Device Registration Certificate. The Xiaosai 1.0 version has been launched, achieving an accuracy rate of approximately 78% in AI mental health assessments and providing effective specialized models for pregnant/postpartum women and workplace populations. Currently, the Xiaosai 2.0 version has undergone further iterative upgrades; when compared with diagnoses by clinical psychiatrists, its accuracy can reach 85%, and it is expected to reduce the average analysis time to 1.5 minutes.

 


B2B-focused, primarily through channel partnerships, with annual revenue exceeding RMB 2 million


In terms of business model, as a technology-driven enterprise, Hongjun Intelligent currentlyB2B-focused, with an emphasis on channel partnerships

 

Notable partners include: the Department of Psychology at a municipal-level Maternal and Child Health Hospital in Shandong Province, where Hongjun Intelligent’s AI assessment tools have been integrated into its internet hospital platform and are utilized in clinical scenarios such as psychiatric ward rounds, outpatient psychology clinics, and psychological counseling at the reproductive center; Yuan Mouyi, a company specializing in the research and development of nervonic acid products primarily targeting consumers (C-end), which has embedded Hongjun Intelligent’s AI assessment tools into its WeChat Mini Program, creating a closed-loop approach to mild mental health interventions by combining assessment software with food and dietary supplement products; and a subsidiary of Zhongke, a technology solutions company focused on mental health that primarily serves hospital psychology departments. This subsidiary has embedded Hongjun Intelligent’s AI assessment software into its intelligent screening system designed for psychology departments, enabling multifaceted mental health services including human-computer interaction. Additionally, Hongjun Intelligent’s AI assessment tools are integrated into the hardware systems provided by this subsidiary to a provincial capital psychiatric hospital, serving both outpatient and community care settings.

 

The company’s revenue model comprises three types.

 

1. Provide SaaS-based psychological health assessments and large-scale mental health management services to large enterprises, internet companies, and other clients, with fees charged per assessment; 2. Due to differing operational business models, enterprises may choose to deploy Hongjun Intelligent’s AI psychological assessment management system on their own servers. After deployment, charges are applied only for the computing power used in assessments and subsequent database upgrades; 3. Enterprises can integrate AI assessment interfaces into their relevant hardware devices. As these devices are sold and promoted, Hongjun Intelligent can also achieve revenue growth. All three models involve high-frequency recurring usage, thereby stabilizing the company’s revenue streams.

 

Under this model,In 2022, Hunan Hongjun Intelligent Technology Co., Ltd. achieved an annual revenue of over RMB 2 million. The company’s revenue is projected to reach RMB 5 million in 2023, with a forecasted total of RMB 50 million over the next three years.

 

 

Onboarded over 180 enterprises and public institutions, focusing on enhancing the accuracy of AI assessments


Hunan Hongjun Intelligent Technology Co., Ltd. was established in 2021. The company’s ability to achieve technological implementation and develop a clear business model within just two years is largely attributable to the efforts of its team.

 

It is understood that the company’s team members hail from Tsinghua University, Sun Yat-sen University, the Second Xiangya Hospital of Central South University, the Institute of Psychology of the Chinese Academy of Sciences, and former Tencent/JD.com affiliates. The Chief Executive Officer, psychological scientists, Chief Technology Officer, and Chief Marketing Officer possess extensive professional theoretical knowledge and business experience in psychology and clinical promotion. The Chief Scientist, Chief Operating Officer, and Chief Technology Officer have substantial practical experience in AI technology development and internet operations.

 

Established two years ago, Hongjun Intelligence has secured 10 software copyrights and 2 patents. It has successfully implemented projects in the fields of education, healthcare, Employee Assistance Programs (EAP), and military-civilian integration. The company now serves over 180 enterprises and public institutions, with its software undergoing more than 100,000 assessments.. Collaborated with big data enterprises and HR management platform companies in cities such as Beijing and Shanghai, achieving highly effective industry market expansion.

 

Hongjun Intelligence also has a clear roadmap for its future development. In the short term, Hongjun Intelligence will focus on improving the accuracy of AI-based psychological assessments and continue to expand its product reach among enterprises, public institutions, hospitals, schools, public safety sectors, Employee Assistance Programs (EAPs), and individual users. Meanwhile, the company aims to establish partnerships with more capital firms and enterprises that share its vision and possess in-depth knowledge of the mental health industry, jointly driving progress for both the company and the industry as a whole.

 

In the long term, Hongjun Intelligent aims to elevate its AI-based psychological recognition technology to the gold standard required for inclusion in clinical guidelines, thereby enabling its deployment in large-scale mental health assessment and screening across China. Once the assessment technology reaches a certain level of maturity, Hongjun Intelligent will enter the digital therapeutics sector, with the goal of providing users with one-stop mental health services.