Machine learning maps which Ghanaian research fields are bleeding talent

Ghana is losing more of its publishing researchers than it gains, and the losses are concentrated in exactly the fields a developing knowledge economy can least afford to empty. A new study in the journal Discover Data mined 14 years of publication records from Scopus and the Web of Science to track, classify and forecast the international movements of Ghanaian research talent, and its numbers point to a widening net outflow that accelerated sharply after 2012.

The researchers, led by Muftawu Hussein with colleagues including Emmanuel Ahene and Issifu Damba Kanzoni, extracted 51,000 publications from 2007 to 2020, 29,000 from Scopus and 22,000 from the Web of Science Core Collection. After removing 12,240 duplicate records that appeared in both databases, defined as pairs sharing the same DOI and at least two other matching metadata fields such as first author, article title, ISSN or publication year, 38,760 publication records remained for analysis. The team focused on researchers affiliated with Ghana who had published at least two works in the window, treating each authorship record as a data point linking a person to an institution and a country.

The time frame was chosen deliberately. Around 2007, many African universities began adopting standardized digital identifiers such as ORCID and Scopus Author ID, which made it far easier to distinguish individual authors and follow their affiliations across institutions and borders. The endpoint of 2020 was set to avoid the distortions of the COVID-19 pandemic, which disrupted academic mobility, funding and publishing worldwide. A 14-year span also gave the analysts enough longitudinal depth to detect major transitions in career trajectories.

Using changes in country affiliation between a researcher’s first and most recent publication, the study sorted scholars into four categories: emigrants, whose academic origin is Ghana and whose final affiliation lies abroad, or who moved between two non-Ghana countries; immigrants, who began outside Ghana and now publish from a Ghanaian affiliation; return migrants, who started in Ghana, left and came back; and non-mobile researchers, whose affiliations stayed in Ghana throughout. The result is stark. Of the 38,760 researchers identified, 58.1 percent, or 22,519, showed evidence of international migration, while 41.9 percent, or 16,240, showed none. Among the mobile group, 45.0 percent, or 17,442, moved out of Ghana, while only 13.1 percent, or 5077, moved in. Of that inbound group, just 5.8 percent began their careers abroad, and 7.3 percent were returnees who had started in Ghana and later come home.

The net migration arithmetic turned negative in 2012 and stayed that way through 2020. Between 2007 and 2011, Ghana actually recorded positive net migration, peaking in 2009 at +26.7 per 1000 researchers, with Health Sciences posting the strongest inflow that year at +13.8, followed by Life Sciences at +6.0, Social Sciences at +4.1, Physical Sciences at +2.0 and multidisciplinary work at +0.8. From 2012 onward the rate flipped, and by 2019 and 2020 it had fallen below −30.0 per 1000 researchers, with 2020 marking the most severe exodus in the series. The authors read this as a considerable outflow of research talent concentrated in the most recent years of the record.

Discipline-level analysis sharpened the picture. Measured by support, the share of transactions containing a given field, Ghana’s greatest losses among emigrants were in Medicine, with 4483 scholars or 25.7 percent, Agricultural and Biological Sciences with 2581 scholars or 14.8 percent, and Social Sciences with 1936 scholars or 11.1 percent. Environmental sciences, Computer science, Biochemistry Genetics and Molecular Biology, Business management and Accounting, and Engineering also exceeded the 3 percent support threshold. At the subject-area level, Health Sciences accounted for the highest loss at 28.9 percent, followed by Life Sciences at 24.1 percent. Yet the same field showed a counter-current: among the 5077 researchers who returned or moved into Ghana, 33.9 percent, or 1721, were in Medicine, and 19.5 percent, or 990, were in Agricultural and Biological Sciences. The authors describe this as a reverse brain gain within an overall brain drain.

The methodological core is association rule mining, a data mining technique borrowed from market basket analysis, adapted here to bibliometric records. Each transaction captured one author’s full publication history from 2007 to 2020, including shifts in country affiliation and field of study. The Apriori algorithm generated frequent itemsets and strong rules above minimum support and confidence thresholds, and each rule was scored on four standard metrics: support, the frequency of the itemset; confidence, the conditional likelihood of the consequent given the antecedent; lift, the ratio of observed co-occurrence to what independence would predict; and conviction, which penalizes rules that would often be wrong under independence. A lift above 1 signals a positive correlation between field and destination; a lift below 1 signals a negative one.

The rules expose where Ghanaian researchers go. The rule Medicine to United States carried a confidence of 0.439, meaning a 43.9 percent likelihood that Ghanaian medical scholars migrate to the United States, with a lift of 1.288 and conviction of 1.175, a moderate positive association. Medicine to the United Kingdom showed only 12.6 percent confidence, though still the sole other destination with significant probability for that field. Computer Science to the United States reached 39.4 percent confidence and Social Sciences to the United States 38.0 percent, with lifts of 1.154 and 1.115 respectively. One rule ran against the grain: Agricultural and Biological Sciences to the United States showed a lift of 0.572 and conviction of 0.818, a negative correlation. On the inbound side, the strongest association was Biochemistry, Genetics and Molecular Biology to the United States, with a lift of 2.729, conviction of 2.056 and 62.5 percent confidence, indicating that many Ghanaian researchers in those areas launched their careers in the United States before appearing in the Ghanaian record. At the country level, the United States and United Kingdom ranked as the most prominent origins, transit points and destinations across emigrants, immigrants and return migrants, with China, South Africa, Germany, Canada and Nigeria also recurring.

For forecasting, the team turned to time series modeling through the PyCaret low-code machine learning environment, concentrating on the two disciplines with the highest emigration rates, Medicine and Agricultural and Biological Sciences. Features were limited to year and field of study, and the data were split 65 percent training and 35 percent testing, with test years falling after the training window to preserve temporal order. Models were trained and evaluated with fivefold cross-validation, and the top performer achieved a mean absolute error of 0.04893 with a root mean square error of similar magnitude. The resulting projections, extended toward 2030, point to continued brain drain pressure in both fields, reinforcing the study’s conclusion that the social sciences, agricultural and biological sciences, and health sciences are the most vulnerable to long-term talent loss.

The authors frame the findings through push–pull, human capital and network theory: underfunded labs and limited career ladders push researchers out, while better facilities, funding and recognition pull them toward established systems, and diaspora and collaboration networks channel the flows along durable routes. They argue that predictive analytics of this kind can give policymakers early warning, supporting targeted retention measures, diaspora engagement and return incentives. The study’s own caveats are worth holding onto. Bibliometric data miss researchers publishing in local or non-indexed journals, author disambiguation based on names and affiliations cannot guarantee complete accuracy, and the models do not probe the personal, institutional or economic motives behind each move, nor anticipate shocks such as policy shifts or global crises. The framework, the authors note, is reproducible for other countries facing similar questions about where their research talent is going.

Subject of Research: Predictive analysis of international migration patterns among Ghanaian researchers using bibliometric data and machine learning

Article Title: Robust predictive analysis of international mobility among research talents

Article References: Hussein, M., Ahene, E., Hassan, A. L., & Kanzoni, I. D. (2026). Robust predictive analysis of international mobility among research talents. Discover Data, 4(1), Article 16. https://doi.org/10.1007/s44248-026-00110-5

Image Credits: AI Generated

DOI: 10.1007/s44248-026-00110-5

Keywords: Ghana, brain drain, researcher mobility, bibliometrics, association rule mining, machine learning, Scopus, Web of Science, net migration rate, talent retention, health sciences, time series forecasting