Spaces:

agenticx
/

TxAgentRAOEval

Sleeping

App Files Files Community

RichardZhu52 commited on May 8

Commit

f6e0f52

1 Parent(s): 6bf89d5

update with secret

Browse files

Files changed (2) hide show

app.py +835 -14
utils.py +13 -4

app.py CHANGED Viewed

@@ -6,6 +6,826 @@ import json
 from utils import format_chat, append_to_sheet, read_sheet_to_df
 import random
 # Load tool lists
 fda_drug_labeling_tools_path = "fda_drug_labeling_tools.json"
 monarch_tools_path = "monarch_tools.json"
@@ -459,20 +1279,21 @@ with gr.Blocks(css=centered_col_css) as demo:
         with gr.Row():
             submit_questions_btn = gr.Button("Submit Questions for TxAgent Evaluation")
             participate_eval_btn = gr.Button("Participate in TxAgent Evaluation")
-        gr.HTML("""
-        <div>
-            <p><strong>About TxAgent:</strong> TxAgent is a first-of-its-kind AI model developed in the Zitnik Lab at Harvard Medical School. It leverages multi-step reasoning and real-time biomedical knowledge retrieval from a toolbox of 211 tools to analyze drug interactions, contraindications, and patient-specific treatment strategies. It is designed to provide personalized treatment recommendations across a wide range of diseases, including rare diseases.</p>
-            <p>TxAgent evaluates how drugs interact at molecular, pharmacokinetic, and clinical levels, identifies contraindications based on patient comorbidities and concurrent medications, and tailors treatment strategies to individual patient characteristics, including age, genetic factors, and disease progression. It achieves 92.1% accuracy in open-ended drug reasoning tasks, surpassing GPT-4o by up to 25.8% and outperforming DeepSeek-R1 (671B) in structured multi-step reasoning.</p>
-            <p>TxAgent's toolbox, ToolUniverse, consolidates 211 tools from trusted sources, including all US FDA-approved drugs since 1939 and validated clinical insights from Open Targets. By integrating multi-step inference, real-time knowledge grounding, and tool-assisted decision-making, TxAgent ensures that treatment recommendations align with established clinical guidelines and real-world evidence, reducing the risk of adverse events and improving therapeutic decision-making.</p>
-            <p>We are currently conducting a user study with physicians, rare disease experts, and others with relevant medical background to assess TxAgent's performance on personalized therapeutic reasoning across multiple criteria, including helpfulness, clinical consensus, and scientific accuracy.</p>
-            <p>Please choose one of the options below:</p>
-            <ul>
-            <li>If you would like to submit new questions for TxAgent to be evaluated on, please click the "Submit Questions" button.</li>
-            <li>If you are here to participate in the evaluation of TxAgent's responses to existing questions, please click the "Participate in Evaluation" button.</li>
-            </ul>
-            <p>Your contribution is invaluable to us. Thank you for your time and expertise!</p>
-        </div>
-        """)
         # Define actions for the new buttons
         # For the Google Form button, we'll use JavaScript to open a new tab.

 from utils import format_chat, append_to_sheet, read_sheet_to_df
 import random
+TxAgent_Project_Page_HTML = """<html>
+  <head>
+    <meta charset="utf-8" />
+    <!-- Meta tags for social media banners, these should be filled in appropriatly as they are your "business card" -->
+    <!-- Replace the content tag with appropriate information -->
+    <meta name="description" content="Project page of TxAgent" />
+    <meta property="og:title" content="TxAgent" />
+    <meta
+      property="og:description"
+      content="An AI Agent for therapeutic reasoning across a universe of tools"
+    />
+    <meta
+      property="og:url"
+      content="https://zitniklab.hms.harvard.edu/TxAgent/"
+    />
+    <!-- Path to banner image, should be in the path listed below. Optimal dimenssions are 1200X630-->
+    <meta property="og:image" content="static/image/TxAgent.jpg" />
+    <meta property="og:image:width" content="1200" />
+    <meta property="og:image:height" content="630" />
+    <meta name="twitter:title" content="TxAgent" />
+    <meta
+      name="twitter:description"
+      content="An AI Agent for therapeutic reasoning across a universe of tools"
+    />
+    <!-- Path to banner image, should be in the path listed below. Optimal dimenssions are 1200X600-->
+    <meta name="twitter:image" content="static/image/TxAgent.jpg" />
+    <meta name="twitter:card" content="summary_large_image" />
+    <!-- Keywords for your paper to be indexed by-->
+    <meta
+      name="keywords"
+      content="AI Agent, Agentic AI, therapeutic reasoning, tool use, tool universe, AI scientist"
+    />
+    <meta name="viewport" content="width=device-width, initial-scale=1" />
+    <title>
+      TxAgent: An AI Agent for Therapeutic Reasoning Across a Universe of Tools
+    </title>
+    <link rel="icon" type="image/x-icon" href="static/images/favicon.ico" />
+    <link
+      href="https://fonts.googleapis.com/css?family=Google+Sans|Noto+Sans|Castoro"
+      rel="stylesheet"
+    />
+    <link rel="stylesheet" href="static/css/bulma.min.css" />
+    <link rel="stylesheet" href="static/css/bulma-carousel.min.css" />
+    <link rel="stylesheet" href="static/css/bulma-slider.min.css" />
+    <link rel="stylesheet" href="static/css/fontawesome.all.min.css" />
+    <link
+      rel="stylesheet"
+      href="https://cdn.jsdelivr.net/gh/jpswalsh/academicons@1/css/academicons.min.css"
+    />
+    <link rel="stylesheet" href="static/css/index.css" />
+    <script src="https://ajax.googleapis.com/ajax/libs/jquery/3.5.1/jquery.min.js"></script>
+    <script src="https://documentcloud.adobe.com/view-sdk/main.js"></script>
+    <script defer src="static/js/fontawesome.all.min.js"></script>
+    <script src="static/js/bulma-carousel.min.js"></script>
+    <script src="static/js/bulma-slider.min.js"></script>
+    <script src="static/js/index.js"></script>
+    <script src="https://code.highcharts.com/highcharts.js"></script>
+    <script src="https://code.highcharts.com/modules/sunburst.js"></script>
+    <script src="https://code.highcharts.com/modules/exporting.js"></script>
+    <style>
+      .tooluniverse-wrapper {
+        display: flex;
+        justify-content: center;
+        align-items: center;
+        width: 100%;
+        height: 100%;
+      }
+      #tooluniverse-container {
+        width: 65%;
+        height: 65%;
+      }
+      .small-text {
+        font-size: 0.65rem;
+      }
+      /* Fixed styles for centering all images */
+      img {
+        display: block;
+        margin-left: auto !important;
+        margin-right: auto !important;
+        max-width: 100%;
+        height: auto;
+      }
+      /* Fix carousel images */
+      .item img {
+        margin: 0 auto !important;
+      }
+      /* Center videos too */
+      video {
+        display: block;
+        margin: 0 auto !important;
+        max-width: 100%;
+      }
+      /* Force responsive image centering */
+      .responsive-image {
+        display: block !important;
+        margin: 0 auto !important;
+        max-width: 90% !important;
+      }
+      /* Simplify content container centering */
+      .content-container {
+        text-align: center;
+      }
+      /* Ensure container centering */
+      .container.is-centered {
+        text-align: center;
+      }
+    </style>
+  </head>
+  <body>
+    <section class="hero">
+      <div class="hero-body">
+        <div class="container is-max-desktop">
+          <div class="columns is-centered">
+            <div class="column has-text-centered">
+              <h1 class="title is-1 publication-title">
+                TxAgent: An AI agent for therapeutic reasoning across a universe
+                of tools
+              </h1>
+              <div class="is-size-5 publication-authors">
+                <!-- Paper authors -->
+                <span class="author-block">
+                  <a href="https://shgao.site" target="_blank">Shanghua Gao</a
+                  ><sup>1</sup>,</span
+                >
+                <span class="author-block">
+                  <a
+                    href="https://www.linkedin.com/in/richard-zhu-4236901a7/"
+                    target="_blank"
+                    >Richard Zhu</a
+                  ><sup>1</sup>,</span
+                >
+                <span class="author-block">
+                  <a href="https://zlkong.github.io/homepage/" target="_blank"
+                    >Zhenglun Kong</a
+                  ><sup>1</sup>,</span
+                >
+                <span class="author-block">
+                  <a href="https://www.ayushnoori.com/" target="_blank"
+                    >Ayush Noori</a
+                  ><sup>1</sup>,</span
+                >
+                <span class="author-block">
+                  <a
+                    href="https://scholar.google.com/citations?hl=zh-CN&user=Awdn73MAAAAJ"
+                    target="_blank"
+                    >Xiaorui Su</a
+                  ><sup>1</sup>,</span
+                >
+                <span class="author-block">
+                  <a
+                    href="https://www.linkedin.com/in/curtisginder/"
+                    target="_blank"
+                    >Curtis Ginder</a
+                  ><sup>1,2</sup>,</span
+                >
+                <span class="author-block">
+                  <a href="https://sites.google.com/view/theo-t" target="_blank"
+                    >Theodoros Tsiligkaridis</a
+                  ><sup>3</sup>,</span
+                >
+                <span class="author-block">
+                  <a href="https://zitniklab.hms.harvard.edu/" target="_blank"
+                    >Marinka Zitnik</a
+                  ><sup>1,4,5,6&Dagger;;</sup></span
+                >
+              </div>
+              <details>
+                <summary>Author Affiliations</summary>
+                <div class="is-size-5 publication-authors">
+                  <span class="author-block"
+                    ><sup>1</sup>Department of Biomedical Informatics, Harvard
+                    Medical School, Boston, MA</span
+                  >
+                  <span class="author-block"
+                    ><sup>2</sup>Cardiovascular Division, Department of
+                    Medicine, Brigham and Women’s Hospital, Harvard Medical
+                    School, Boston, MA</span
+                  >
+                  <span class="author-block"
+                    ><sup>3</sup>MIT Lincoln Laboratory, Lexington, MA</span
+                  >
+                  <span class="author-block"
+                    ><sup>4</sup>Kempner Institute for the Study of Natural and
+                    Artificial Intelligence, Harvard University, Cambridge,
+                    MA</span
+                  >
+                  <span class="author-block"
+                    ><sup>5</sup>Broad Institute of MIT and Harvard, Cambridge,
+                    MA</span
+                  ><br />
+                  <span class="author-block"
+                    ><sup>6</sup>Harvard Data Science Initiative, Cambridge,
+                    MA</span
+                  >
+                </div>
+              </details>
+              <div class="column has-text-centered">
+                <div class="publication-links">
+                  <!-- Arxiv PDF link -->
+                  <span class="link-block">
+                    <a
+                      href="https://arxiv.org/pdf/2503.10970"
+                      target="_blank"
+                      class="external-link button is-normal is-rounded is-dark"
+                    >
+                      <span class="icon">
+                        <i class="fas fa-file-pdf"></i>
+                      </span>
+                      <span>Paper</span>
+                    </a>
+                  </span>
+                  <!-- Github link -->
+                  <span class="link-block">
+                    <a
+                      href="https://github.com/mims-harvard/TxAgent"
+                      target="_blank"
+                      class="external-link button is-normal is-rounded is-dark"
+                    >
+                      <span class="icon">
+                        <i class="fab fa-github"></i>
+                      </span>
+                      <span>TxAgent code</span>
+                    </a>
+                  </span>
+                  <span class="link-block">
+                    <a
+                      href="https://github.com/mims-harvard/ToolUniverse"
+                      target="_blank"
+                      class="external-link button is-normal is-rounded is-dark"
+                    >
+                      <span class="icon">
+                        <i class="fab fa-github"></i>
+                      </span>
+                      <span>ToolUniverse code</span>
+                    </a>
+                  </span>
+                  <!-- Github link -->
+                  <span class="link-block">
+                    <a
+                      href="https://huggingface.co/collections/mims-harvard/txagent-67c8e54a9d03a429bb0c622c"
+                      target="_blank"
+                      class="external-link button is-normal is-rounded is-dark"
+                    >
+                      <span class="icon"> 🤗 </span>
+                      <span>HuggingFace</span>
+                    </a>
+                  </span>
+                  <span class="link-block">
+                    <a
+                      href="https://kempnerinstitute.harvard.edu/research/deeper-learning/txagent-an-ai-agent-for-therapeutic-reasoning-across-a-universe-of-211-tools/"
+                      target="_blank"
+                      class="external-link button is-normal is-rounded is-dark"
+                    >
+                      <span>Kempner Blog</span>
+                    </a>
+                  </span>
+                </div>
+              </div>
+            </div>
+          </div>
+        </div>
+      </div>
+    </section>
+    <!-- Model overview -->
+    <section class="hero is-small">
+      <div class="hero-body content-container">
+        <div class="container">
+          <h2 class="title is-2 has-text-centered">TxAgent model</h2>
+          <div class="container is-centered has-text-centered">
+            <img
+              src="static/images/txagent.jpg"
+              alt="TxAgent model"
+              style="height: auto; max-width: 48vw"
+              class="responsive-image"
+            />
+          </div>
+        </div>
+      </div>
+    </section>
+    <!-- End model overview -->
+    <!-- Paper abstract -->
+    <section class="section hero is-light">
+      <div class="container is-max-desktop">
+        <div class="columns is-centered has-text-centered">
+          <div class="column is-four-fifths">
+            <h2 class="title is-3">Abstract</h2>
+            <div class="content has-text-justified">
+              <p>
+                Precision therapeutics require multimodal adaptive models that
+                generate personalized treatment recommendations. We introduce
+                TxAgent, an AI agent that leverages multi-step reasoning and
+                real-time biomedical knowledge retrieval across a toolbox of 211
+                tools to analyze drug interactions, contraindications, and
+                patient-specific treatment strategies. TxAgent evaluates how
+                drugs interact at molecular, pharmacokinetic, and clinical
+                levels, identifies contraindications based on patient
+                comorbidities and concurrent medications, and tailors treatment
+                strategies to individual patient characteristics, including age,
+                genetic factors, and disease progression. TxAgent retrieves and
+                synthesizes evidence from multiple biomedical sources, assesses
+                interactions between drugs and patient conditions, and refines
+                treatment recommendations through iterative reasoning. It
+                selects tools based on task objectives and executes structured
+                function calls to solve therapeutic tasks that require clinical
+                reasoning and cross-source validation. The ToolUniverse
+                consolidates 211 tools from trusted sources, including all US
+                FDA-approved drugs since 1939 and validated clinical insights
+                from Open Targets. TxAgent outperforms leading LLMs, tool-use
+                models, and reasoning agents across five new benchmarks: DrugPC,
+                BrandPC, GenericPC, TreatmentPC, and DescriptionPC, covering
+                3,168 drug reasoning tasks and 456 personalized treatment
+                scenarios. It achieves 92.1% accuracy in open-ended drug
+                reasoning tasks, surpassing GPT-4o by up to 25.8% and
+                outperforming DeepSeek-R1 (671B) in structured multi-step
+                reasoning. TxAgent generalizes across drug name variants and
+                descriptions, maintaining a variance of &lt;0.01 between brand,
+                generic, and description-based drug references, exceeding
+                existing tool-use LLMs by over 55%. By integrating multi-step
+                inference, real-time knowledge grounding, and tool- assisted
+                decision-making, TxAgent ensures that treatment recommendations
+                align with established clinical guidelines and real-world
+                evidence, reducing the risk of adverse events and improving
+                therapeutic decision-making.
+              </p>
+            </div>
+          </div>
+        </div>
+      </div>
+    </section>
+    <!-- End paper abstract -->
+    <!-- Use TxAgent -->
+    <section class="hero is-small">
+      <div class="hero-body content-container">
+        <h2 class="title is-2 has-text-centered py-2">A Simple Guide to Using TxAgent</h2>
+        <div class="container is-centered has-text-centered is-max-desktop">
+          <section class="section hero is-light py-2">
+            <div class="hero-body has-text-left py-2">
+              <h2 class="has-text-weight-normal">
+                Install ToolUniverse:
+                <code> pip install tooluniverse </code>
+                <br />
+                Install TxAgent:
+                <code> pip install txagent </code>
+                <br />
+                Run TxAgent demo/script:
+                <code> https://github.com/mims-harvard/TxAgent </code>
+            </div>
+          </section>
+        </div>
+      </div>
+    </section>
+    <!-- End use txagent -->
+    <!-- Model capabilities -->
+    <section class="hero is-small">
+      <div class="hero-body content-container">
+        <div class="container">
+          <h2 class="title is-2 has-text-centered">TxAgent capabilities</h2>
+          <div class="container is-centered has-text-centered">
+            <img
+              src="static/images/txagent_capabilities.jpg"
+              alt="TxAgent Capabilities"
+              style="height: auto; max-width: 50vw"
+              class="responsive-image"
+            />
+          </div>
+          <div class="container is-centered has-text-centered is-max-desktop">
+            <section class="section hero is-light py-2 content">
+              <div class="has-text-left py-2">
+                <ul>
+                  <li>
+                    <b>Knowledge grounding using tool calls</b>: TxAgent
+                    utilizes tools to obtain verified knowledge and provides
+                    outputs based on it.
+                  </li>
+                  <li>
+                    <b>Goal-oriented tool selection</b>: TxAgent proactively
+                    requests tools from ToolUniverse using the ToolRAG model and
+                    selects and applies the most suitable tool from the
+                    available candidates.
+                  </li>
+                  <li>
+                    <b>Problem solving with multi-step reasoning</b>: TxAgent
+                    manages complex tasks or unexpected responses from tools
+                    through multiple iterations of thought and function calls.
+                  </li>
+                  <li>
+                    <b>Leveraging constantly updated knowledge bases</b>:
+                    TxAgent accesses continuously updated databases via tools to
+                    handle problems that go beyond the TxAgent’s intrinsic
+                    knowledge.
+                  </li>
+                </ul>
+              </div>
+            </section>
+          </div>
+        </div>
+      </div>
+    </section>
+    <!-- End model capabilities -->
+    <!-- Tool universe -->
+    <section class="hero is-small">
+      <div class="hero-body content-container">
+        <h2 class="title is-2 has-text-centered py-2">ToolUniverse</h2>
+        <!-- Image carousel -->
+        <section class="hero is-small">
+          <div class="hero-body">
+            <div class="container has-text-centered">
+              <div id="results-carousel" class="carousel results-carousel">
+                <div class="item">
+                  <!-- Your image here -->
+                  <img
+                    src="static/images/tool_openfda.jpg"
+                    alt="MY ALT TEXT"
+                    style="height: auto; max-width: 50vw"
+                  />
+                  <h2 class="subtitle has-text-centered">
+                    Tool example linked to OpenFDA.
+                  </h2>
+                </div>
+                <div class="item">
+                  <!-- Your image here -->
+                  <img
+                    src="static/images/tool_opentarget.jpg"
+                    alt="MY ALT TEXT"
+                    style="height: auto; max-width: 50vw"
+                  />
+                  <h2 class="subtitle has-text-centered">
+                    Tool example linked to OpenTarget.
+                  </h2>
+                </div>
+                <div class="item">
+                  <!-- Your image here -->
+                  <img
+                    src="static/images/tool_monarch.jpg"
+                    alt="MY ALT TEXT"
+                    style="height: auto; max-width: 50vw"
+                  />
+                  <h2 class="subtitle has-text-centered">
+                    Tool example linked to Monarch Initiative.
+                  </h2>
+                </div>
+              </div>
+            </div>
+          </div>
+        </section>
+        <!-- End image carousel -->
+        <div class="container is-centered has-text-centered is-max-desktop">
+          <section class="section hero is-light py-2">
+            <div class="hero-body has-text-left py-2">
+              <h2 class="has-text-weight-normal">
+                ToolUniverse is a critical component of TxAgent, providing the
+                agent with the ability to access and leverage a vast array of
+                biomedical knowledge to solve complex therapeutic reasoning
+                tasks. ToolUniverse includes 211 biomedical tools that address
+                various aspects of drugs and diseases. These tools are linked to
+                trusted sources, including all US FDA-approved drugs since 1939
+                and validated clinical insights from Open Targets and Monarch
+                Initiative.
+                <br />
+                Install
+                <a href="https://github.com/mims-harvard/ToolUniverse/"
+                  >ToolUniverse</a
+                >
+                with one line of code:
+                <code> pip install tooluniverse </code>
+              </h2>
+            </div>
+          </section>
+        </div>
+        <div class="tooluniverse-wrapper">
+          <div id="tooluniverse-container"></div>
+        </div>
+      </div>
+    </section>
+    <!-- End tool universe -->
+    <!-- Model training -->
+    <section class="hero is-small">
+      <div class="hero-body content-container">
+        <div class="container">
+          <h2 class="title is-2 has-text-centered">
+            Training TxAgent: TxAgent-instruct dataset
+          </h2>
+          <div class="container is-centered has-text-centered">
+            <img
+              src="static/images/txagent_train.jpg"
+              alt="TxAgent Capabilities"
+              style="height: auto; max-width: 50vw"
+              class="responsive-image"
+            />
+          </div>
+          <div class="container is-centered has-text-centered is-max-desktop">
+            <section class="section hero is-light py-2">
+              <div class="hero-body has-text-left py-2">
+                <h2 class="has-text-weight-normal">
+                  The multi-agent systems (i.e., ToolGen, QuestionGen, and
+                  TraceGen) construct the TxAgent-Instruct training dataset for
+                  instruction tuning LLM to achieve the capabilities of TxAgent.
+                  TxAgent-Instruct is a diverse, synthetic, multi-step reasoning
+                  and large-scale function call training dataset anchored in
+                  biomedical knowledge. It consists of 378,027
+                  instruction-tuning data samples, with each sample generated by
+                  breaking down complete reasoning traces into step-by-step
+                  training data. These samples are derived from 85,340
+                  multi-step reasoning traces, which collectively include
+                  177,626 reasoning steps and 281,695 function calls.
+                </h2>
+              </div>
+            </section>
+          </div>
+        </div>
+      </div>
+    </section>
+    <!-- End model training -->
+    <!-- Multi-agent system -->
+    <section class="hero is-small">
+      <div class="hero-body content-container">
+        <div class="container">
+          <h2 class="title is-2 has-text-centered">
+            Building TxAgent-Instruct: multi-agent systems
+          </h2>
+          <div class="container is-centered has-text-centered">
+            <img
+              src="static/images/multi-agent.jpg"
+              alt="TxAgent Capabilities"
+              style="height: auto; max-width: 50vw"
+              class="responsive-image"
+            />
+          </div>
+          <div class="container is-centered has-text-centered is-max-desktop">
+            <section class="section hero is-light py-2 content">
+              <div class="has-text-left py-2">
+                <ul>
+                  <li>
+                    <b>ToolGen</b>: A tool generation multi-agent system that
+                    transforms APIs into 211 agent-compatible tools, aggregating
+                    them into the ToolUniverse.
+                  </li>
+                  <li>
+                    <b>QuestionGen</b>: A question generation multi-agent system
+                    designed to extract critical information from documents
+                    (e.g., FDA drug documentation) and generate relevant
+                    questions.
+                  </li>
+                  <li>
+                    <b>TraceGen</b>: A reasoning trace generation multi-agent
+                    system, where a Helper agent and a Tool Provider module
+                    assist the Solver agent in generating step-by-step reasoning
+                    and function calls to solve a problem.
+                  </li>
+                </ul>
+              </div>
+            </section>
+          </div>
+        </div>
+      </div>
+    </section>
+    <!-- End multi-agent system -->
+    <!-- DrugPC -->
+    <section class="hero is-small">
+      <div class="hero-body content-container">
+        <div class="container">
+          <h2 class="title is-2 has-text-centered">
+            Drug reasoning on 11 tasks
+          </h2>
+          <div class="container is-centered has-text-centered">
+            <img
+              src="static/images/drugpc-open-end.jpg"
+              alt="TxAgent Capabilities"
+              style="height: auto; max-width: 50vw"
+              class="responsive-image"
+            />
+          </div>
+          <div class="container is-centered has-text-centered is-max-desktop">
+            <section class="section hero is-light py-2">
+              <div class="hero-body has-text-left py-2">
+                <h2 class="has-text-weight-normal">
+                  DrugPC (Drug Prescribing Card) benchmark includes 3,168
+                  questions covering 11 common tasks related to drugs. TxAgent
+                  outperforms larger open-source LLMs and GPT-4 as well as
+                  existing tool-use LLMs across 11 tasks, excelling in both
+                  open-ended questions. These tasks cover various drug-related
+                  topics, including drug overview, ingredients, warnings and
+                  safety, dependence and abuse, dosage and administration, use
+                  in specific populations, pharmacology, clinical information,
+                  nonclinical toxicology, patient-focused information, and
+                  storage and supply.
+                </h2>
+              </div>
+            </section>
+          </div>
+        </div>
+      </div>
+    </section>
+    <!-- End DrugPC -->
+    <!-- TreatmentPC -->
+    <section class="hero is-small">
+      <div class="hero-body content-container">
+        <div class="container">
+          <h2 class="title is-2 has-text-centered">
+            Specialized treatment recommendations
+          </h2>
+          <div class="container is-centered has-text-centered">
+            <img
+              src="static/images/treatmentpc.jpg"
+              alt="TxAgent Capabilities"
+              style="height: auto; max-width: 50vw"
+              class="responsive-image"
+            />
+          </div>
+          <div class="container is-centered has-text-centered is-max-desktop">
+            <section class="section hero is-light py-2">
+              <div class="hero-body has-text-left py-2">
+                <h2 class="has-text-weight-normal">
+                  TreatmentPC (Treatment Prescribing Cards) benchmark includes
+                  456 questions regarding specialized treatment recommendations.
+                  While multiple indications can be applied to a single disease,
+                  patients with specific conditions—such as pregnancy or
+                  comorbidities—require specialized treatment approaches,
+                  including customized drug selection and dosage adjustments.
+                  The TreatmentPC benchmark is designed to evaluate such
+                  specialized treatment scenarios with questions that account
+                  for the varying application conditions of drugs.
+                  <br />
+                  TxAgnet outperforms larger LLMs such as GPT-4o and Llama
+                  3.1-70B-Instruct as well as tool-use LLMs in both open-ended
+                  and multiple-choice settings. TxAgent achieves superior
+                  performance compared to the full DeepSeek-R1 model (671B) and
+                  its two distilled versions based on Llama-3.1-8B and
+                  Llama-3.3-70B.
+                </h2>
+              </div>
+            </section>
+          </div>
+        </div>
+      </div>
+    </section>
+    <!-- End TreatmentPC -->
+    <!-- Demo -->
+    <section class="hero is-small">
+      <div class="hero-body content-container">
+        <h2 class="title is-2 has-text-centered py-2">TxAgent demos</h2>
+        <!-- Image carousel -->
+        <section class="hero is-small">
+          <div class="hero-body">
+            <div class="container has-text-centered">
+              <div id="results-carousel" class="carousel results-carousel">
+                <div class="item">
+                  <video
+                    poster=""
+                    id="tree"
+                    autoplay
+                    controls
+                    muted
+                    loop
+                    height="100%"
+                  >
+                    <!-- Your video here -->
+                    <source src="static/videos/q1.mov" type="video/mp4" />
+                  </video>
+                  <h2 class="subtitle has-text-centered">
+                    Demo 1: Treatment adjustments considering drug mechanism and
+                    patient condition.
+                  </h2>
+                </div>
+                <div class="item">
+                  <video
+                    poster=""
+                    id="tree"
+                    autoplay
+                    controls
+                    muted
+                    loop
+                    height="100%"
+                  >
+                    <!-- Your video here -->
+                    <source src="static/videos/q2.mov" type="video/mp4" />
+                  </video>
+                  <h2 class="subtitle has-text-centered">
+                    Demo2: Treatment adjustments considering the precise patient
+                    condition (e.g., changing the patient condition from
+                    moderate to severe hepatic impairment).
+                  </h2>
+                </div>
+                <div class="item">
+                  <video
+                    poster=""
+                    id="tree"
+                    autoplay
+                    controls
+                    muted
+                    loop
+                    height="100%"
+                  >
+                    <!-- Your video here -->
+                    <source src="static/videos/q3.mov" type="video/mp4" />
+                  </video>
+                  <h2 class="subtitle has-text-centered">
+                    Demo 3: Treatment selection considering drug-drug
+                    interactions.
+                  </h2>
+                </div>
+              </div>
+            </div>
+          </div>
+        </section>
+        <!-- End image carousel -->
+        <div class="container is-centered has-text-centered is-max-desktop">
+          <section class="section hero is-light py-2">
+            <div class="hero-body has-text-left py-2">
+              <h2 class="has-text-weight-normal">
+                Get code in
+                <a href="https://github.com/mims-harvard/TxAgent/">TxAgent</a>!
+                Lauch the TxAgent demo to see how TxAgent can assist in
+                therapeutic reasoning across a universe of tools.
+              </h2>
+            </div>
+          </section>
+        </div>
+        <div class="tooluniverse-wrapper">
+          <div id="tooluniverse-container"></div>
+        </div>
+      </div>
+    </section>
+    <!-- End Demo -->
+    <!--BibTex citation -->
+    <section class="section" id="BibTeX">
+      <div class="container is-max-desktop content">
+        <h2 class="title">BibTeX</h2>
+        <pre><code>@misc{gao2025txagent,
+          title={TxAgent: An AI Agent for Therapeutic Reasoning Across a Universe of Tools},
+          author={Shanghua Gao and Richard Zhu and Zhenglun Kong and Ayush Noori and Xiaorui Su and Curtis Ginder and Theodoros Tsiligkaridis and Marinka Zitnik},
+          year={2025},
+          eprint={2503.10970},
+          archivePrefix={arXiv},
+          primaryClass={cs.AI},
+          url={https://arxiv.org/abs/2503.10970},
+    }</code></pre>
+      </div>
+    </section>
+    <!--End BibTex citation -->
+        <!--BibTex citation -->
+        <section class="section" id="BibTeX">
+          <div class="container is-max-desktop content">
+            <h2 class="title">Contact</h2>
+            <p>If you have any questions or suggestions, please email
+              <a href="mailto:[email protected]">Shanghua Gao</a> and
+              <a href="mailto:[email protected]">Marinka Zitnik</a>.
+            </p>
+          </div>
+        </section>
+        <!--End BibTex citation -->
+    <footer class="footer small-text">
+      <div class="container">
+        <div class="columns is-centered">
+          <div class="column is-8">
+            <div class="content">
+              <p>
+                We gratefully acknowledge the support of NIH R01-HD108794, NSF CAREER 2339524, US DoD FA8702-15-D-0001, Harvard Data Science Initiative, Amazon Faculty Research, Google Research Scholar Program, AstraZeneca Research, Roche Alliance with Distinguished Scientists, Sanofi iDEA-iTECH, Pfizer Research, Gates Foundation (INV-079038), Chan Zuckerberg Initiative, John and Virginia Kaneb Fellowship at Harvard Medical School, Biswas Computational Biology Initiative in partnership with the Milken Institute, Harvard Medical School Dean's Innovation Fund for the Use of Artificial Intelligence, and Kempner Institute for the Study of Natural and Artificial Intelligence at Harvard University.  Any opinions, findings, conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the funders.
+                We thank Owen Queen and Thomas Hartvigsen for their valuable discussions on this project and NVIDIA AI for providing access to DeepSeek R1 models.
+              <p>
+                This page was built using the
+                <a
+                  href="https://github.com/eliahuhorwitz/Academic-project-page-template"
+                  target="_blank"
+                  >Academic Project Page Template</a
+                >
+                which was adopted from the <a
+                  href="https://nerfies.github.io"
+                  target="_blank"
+                  >Nerfies</a
+                > project page. You are free to borrow the source code of this
+                website, we just ask that you link back to this page in the
+                footer.
+                This website is licensed under a
+                <a
+                  rel="license"
+                  href="http://creativecommons.org/licenses/by-sa/4.0/"
+                  target="_blank"
+                  >Creative Commons Attribution-ShareAlike 4.0 International
+                  License</a
+                >.
+              </p>
+            </div>
+          </div>
+        </div>
+      </div>
+    </footer>
+  </body>
+</html>"""
 # Load tool lists
 fda_drug_labeling_tools_path = "fda_drug_labeling_tools.json"
 monarch_tools_path = "monarch_tools.json"
         with gr.Row():
             submit_questions_btn = gr.Button("Submit Questions for TxAgent Evaluation")
             participate_eval_btn = gr.Button("Participate in TxAgent Evaluation")
+        gr.HTML(TxAgent_Project_Page_HTML)
+        # gr.HTML("""
+        # <div>
+        #     <p><strong>About TxAgent:</strong> TxAgent is a first-of-its-kind AI model developed in the Zitnik Lab at Harvard Medical School. It leverages multi-step reasoning and real-time biomedical knowledge retrieval from a toolbox of 211 tools to analyze drug interactions, contraindications, and patient-specific treatment strategies. It is designed to provide personalized treatment recommendations across a wide range of diseases, including rare diseases.</p>
+        #     <p>TxAgent evaluates how drugs interact at molecular, pharmacokinetic, and clinical levels, identifies contraindications based on patient comorbidities and concurrent medications, and tailors treatment strategies to individual patient characteristics, including age, genetic factors, and disease progression. It achieves 92.1% accuracy in open-ended drug reasoning tasks, surpassing GPT-4o by up to 25.8% and outperforming DeepSeek-R1 (671B) in structured multi-step reasoning.</p>
+        #     <p>TxAgent's toolbox, ToolUniverse, consolidates 211 tools from trusted sources, including all US FDA-approved drugs since 1939 and validated clinical insights from Open Targets. By integrating multi-step inference, real-time knowledge grounding, and tool-assisted decision-making, TxAgent ensures that treatment recommendations align with established clinical guidelines and real-world evidence, reducing the risk of adverse events and improving therapeutic decision-making.</p>
+        #     <p>We are currently conducting a user study with physicians, rare disease experts, and others with relevant medical background to assess TxAgent's performance on personalized therapeutic reasoning across multiple criteria, including helpfulness, clinical consensus, and scientific accuracy.</p>
+        #     <p>Please choose one of the options below:</p>
+        #     <ul>
+        #     <li>If you would like to submit new questions for TxAgent to be evaluated on, please click the "Submit Questions" button.</li>
+        #     <li>If you are here to participate in the evaluation of TxAgent's responses to existing questions, please click the "Participate in Evaluation" button.</li>
+        #     </ul>
+        #     <p>Your contribution is invaluable to us. Thank you for your time and expertise!</p>
+        # </div>
+        # """)
         # Define actions for the new buttons
         # For the Google Form button, we'll use JavaScript to open a new tab.

utils.py CHANGED Viewed

@@ -11,10 +11,19 @@ import json
 import gradio as gr
 import os
-GSERVICE_ACCOUNT_INFO = os.environ.get("GSERVICE_ACCOUNT_INFO")
-# Optionally, get the sheet name from environment (or use a default)
-# GSHEET_NAME = os.environ.get("GSHEET_NAME", "Your Google Sheet Name")
 GSHEET_NAME = "TxAgent_data_collection"
 #Exponential backoff retry decorator

 import gradio as gr
 import os
+GSERVICE_ACCOUNT_INFO = {
+  "type": "service_account",
+  "project_id": "txagent",
+  "private_key_id": "cc1a12e427917244a93faf6f19e72b589a685e65",
+  "private_key": os.getenv("GSheets_Shanghua_PrivateKey"),
+  "client_email": "[email protected]",
+  "client_id": "108950722202634464257",
+  "auth_uri": "https://accounts.google.com/o/oauth2/auth",
+  "token_uri": "https://oauth2.googleapis.com/token",
+  "auth_provider_x509_cert_url": "https://www.googleapis.com/oauth2/v1/certs",
+  "client_x509_cert_url": "https://www.googleapis.com/robot/v1/metadata/x509/shanghua%40txagent.iam.gserviceaccount.com",
+  "universe_domain": "googleapis.com"
+}
 GSHEET_NAME = "TxAgent_data_collection"
 #Exponential backoff retry decorator