<!DOCTYPE article PUBLIC "-//NLM//DTD JATS (Z39.96) Journal Archiving and Interchange DTD with MathML3 v1.3 20210610//EN" "JATS-archivearticle1-3-mathml3.dtd"><article xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"
  dtd-version="1.3" xml:lang="en" article-type="research-article">
  <?DTDIdentifier.IdentifierValue -//NLM//DTD JATS (Z39.96) Journal Publishing DTD v1.2 20190208//EN?>
  <?DTDIdentifier.IdentifierType public?>
  <?SourceDTD.DTDName JATS-journalpublishing1.dtd?>
  <?SourceDTD.Version 1.2?>
  <?ConverterInfo.XSLTName jats2jats3.xsl?>
  <?ConverterInfo.Version 1?>
  <?properties open_access?>
  <front>
    <journal-meta>
      <journal-id journal-id-type="iso-abbrev">Pharmacophore</journal-id>
      <journal-id journal-id-type="publisher-id">pharmacophorejournal.com</journal-id>
      <journal-id journal-id-type="publisher-id">Pharmacophore</journal-id>
      <journal-title-group>
        <journal-title>Pharmacophore</journal-title>
      </journal-title-group>
      <issn pub-type="epub">2229-5402</issn>
    </journal-meta>
    <article-meta>
      <article-id pub-id-type="publisher-id">pharmacophorejournal.com-6991</article-id>
      <article-id pub-id-type="doi">10.51847/o6iDrAOuiq</article-id>
      <article-categories>
        <subj-group subj-group-type="heading">
          <subject>Original research</subject>
        </subj-group>
      </article-categories>
      <title-group>
        <article-title>Connecting Molecular Design with Patient-Level Treatment Decisions: An Integrative Review of Multimodal Foundation Models, Cross-Scale Biomedical Representations, and Evidence Gaps in Pharmaceutical and Precision Medicine Applications</article-title>
      </title-group>
                    <contrib-group>
                      <contrib contrib-type="author">
              <name>
                <surname>Tan</surname>
                <given-names>Michael</given-names>
              </name>
                              <xref rid="aff1" ref-type="aff">1</xref>
                                                            <xref rid="cor1" ref-type="corresp" />
                          </contrib>
                      <contrib contrib-type="author">
              <name>
                <surname>Aziz</surname>
                <given-names>Farah</given-names>
              </name>
                              <xref rid="aff2" ref-type="aff">2</xref>
                                        </contrib>
                      <contrib contrib-type="author">
              <name>
                <surname>Nagy</surname>
                <given-names>Peter</given-names>
              </name>
                              <xref rid="aff3" ref-type="aff">3</xref>
                                        </contrib>
                      <contrib contrib-type="author">
              <name>
                <surname>de Vries</surname>
                <given-names>Johan</given-names>
              </name>
                              <xref rid="aff4" ref-type="aff">4</xref>
                                        </contrib>
                  </contrib-group>
                  <aff id="aff1">
            <label>1</label>Department of Pharmaceutical AI and Precision Medicine, National University of Singapore, Singapore.
          </aff>
                  <aff id="aff2">
            <label>2</label>Department of Multimodal Biomedical Data Science, Universiti Putra Malaysia, Serdang, Malaysia.
          </aff>
                  <aff id="aff3">
            <label>3</label>Department of Pharmaceutical Data Science and Foundation Models, Corvinus University of Budapest, Budapest, Hungary.
          </aff>
                  <aff id="aff4">
            <label>4</label>Department of Precision Medicine and Cross-Scale Biomedical Modeling, Erasmus University Rotterdam, Rotterdam, Netherlands.
          </aff>
                          <author-notes>
            <corresp id="cor1">
              <bold>Address for correspondence:</bold> Prof. Wael Abu Dayyih, Department of
              Pharmaceutical Chemistry, Faculty of Pharmacy, Mutah University, Al-Karak 61710, Jordan.
                              E-mail: <email xlink:href="michael.tan@nus.edu.sg">michael.tan@nus.edu.sg</email>
                          </corresp>
          </author-notes>
                    <pub-date pub-type="epub">
        <day>28</day>
        <month>10</month>
        <year>2026</year>
      </pub-date>
      <volume>17</volume>
      <issue>5</issue>
      <fpage>82</fpage>
      <lpage>91</lpage>
      <permissions>
        <copyright-statement>
          Copyright: &#x000a9; 2026 Pharmacophore
        </copyright-statement>
        <copyright-year>2026</copyright-year>
        <license>
          <ali:license_ref xmlns:ali="http://www.niso.org/schemas/ali/1.0/"
            specific-use="textmining" content-type="ccbyncsalicense">
            https://creativecommons.org/licenses/by-nc-sa/4.0/</ali:license_ref>
          <license-p>This is an open access journal, and articles are distributed under the terms of
            the Creative Commons Attribution-NonCommercial-ShareAlike 4.0 License, which allows
            others to remix, tweak, and build upon the work non-commercially, as long as appropriate
            credit is given and the new creations are licensed under the identical terms.</license-p>
        </license>
      </permissions>
      <abstract>
        <title>A<sc>BSTRACT</sc></title>
        <p>Multimodal foundation models are expanding the range of biomedical information that artificial intelligence can represent, combining molecular structures, omic profiles, cellular states, medical images, and clinical records within increasingly general computational systems. However, the capacity to represent heterogeneous information does not establish that molecular design outputs can inform individualized treatment decisions. This integrative review examines the evidentiary connections between pharmaceutical discovery and precision-medicine applications, distinguishing computational representation, biological validation, clinical prediction, and treatment selection. The review combines molecular-design studies, biomedical foundation-model evaluations, multimodal integration research, pharmacological response models, and clinically oriented investigations. A bounded bibliographic search and structured evidence mapping inform the synthesis; complete database-wide screening and full-text inclusion denominators remain unresolved. Evidence is organized according to biomedical scale, representation type, validation setting, and the specific decision supported. The central contribution is a cross-scale interpretation that distinguishes directly evaluated connections from relationships inferred through adjacent studies and transitions that remain untested. Molecular and cellular models provide increasingly sophisticated representations, while clinical investigations demonstrate selected routes from patient profiling to treatment matching. Evidence connecting these capabilities within a single prospectively validated molecular-design-to-treatment system remains more limited.  The review identifies the methodological and translational conditions under which multimodal representations could become clinically actionable, emphasizing biological context, independent validation, comparative treatment evidence, and preservation of uncertainty across successive scales.</p>
      </abstract>
      <kwd-group>
                <kwd>Multimodal foundation models</kwd>
                <kwd>Molecular design</kwd>
                <kwd>Multi-omics integration</kwd>
                <kwd>Drug response prediction</kwd>
                <kwd>Precision medicine</kwd>
                <kwd>Cross-scale biomedical representations</kwd>
              </kwd-group>
    </article-meta>
  </front>
</article>
